奉献
Dedication
致我的妻子悉尼
To my wife, Sydney
题词
Epigraph
在遥远的未来，我看到了更为重要的研究领域。心理学将建立在一个新的基础之上，即每一种心理力量和能力的获得都是循序渐进的。人类的起源及其历史将得到揭示。
In the distant future I see open fields for far more important researches. Psychology will be based on a new foundation, that of the necessary acquirement of each mental power and capacity by gradation. Light will be thrown on the origin of man and his history.
—查尔斯·达尔文， 1859 年
—CHARLES DARWIN IN 1859
内容
Contents
The Basics of Human Brain Anatomy
Breakthrough #1: Steering and the First Bilaterians
4: Associating, Predicting, and the Dawn of Learning
Breakthrough #2: Reinforcing and the First Vertebrates
6: The Evolution of Temporal Difference Learning
7: The Problems of Pattern Recognition
9: The First Model of the World
Breakthrough #3: Simulating and the First Mammals
11: Generative Models and the Neocortical Mystery
13: Model-Based Reinforcement Learning
14: The Secret to Dishwashing Robots
Breakthrough #4: Mentalizing and the First Primates
15: The Arms Race for Political Savvy
17: Monkey Hammers and Self-Driving Cars
18: Why Rats Can’t Go Grocery Shopping
Breakthrough #5: Speaking and the First Humans
19: The Search for Human Uniqueness
22: ChatGPT and the Window into the Mind
Conclusion: The Sixth Breakthrough
人类大脑解剖学基础知识
The Basics of Human Brain Anatomy
原创艺术由 Mesa Schumacher 创作
Original art by Mesa Schumacher
我们的进化谱系
Our Evolutionary Lineage
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
特别感谢 Rebecca Gelernter 为本书创作了令人难以置信的原创艺术作品；Rebecca 创作了每个突破部分开头的艺术作品，并设计了大部分人物形象。此外，还要特别感谢 Mesa Schumacher 为本书专门创作的精彩原创人体、七鳃鳗、猴子和老鼠大脑解剖艺术作品。
Special thanks to Rebecca Gelernter for creating the incredible original art in this book; Rebecca created the art at the beginning of each Breakthrough section and designed the majority of the figures. Also, a special thanks to Mesa Schumacher for her wonderful original anatomical art of the human, lamprey, monkey, and rat brain that she made specifically for this book.
介绍
Introduction
1962 年 9 月，在全球太空竞赛、古巴导弹危机和新近升级的脊髓灰质炎疫苗的动荡之中，出现了一个较少被报道但或许同样重要的人类历史上的里程碑：正是在 1962 年秋天，我们预测了未来。
IN SEPTEMBER 1962, during the global tumult of the space race, the Cuban missile crisis, and the recently upgraded polio vaccine, there was a less reported—but perhaps equally critical—milestone in human history: It was in the fall of ’62 that we predicted the future.
在美国电视机新推出的彩色屏幕上，首次播出了《杰森一家》这部动画片，讲述的是一个生活在一百年后的家庭。这部动画片以情景喜剧的形式，实际上预测了未来人类的生活，以及哪些技术会填满他们的口袋、装饰他们的家。
Cast onto the newly colorful screens of American televisions was the debut of The Jetsons, a cartoon about a family living one hundred years in the future. In the guise of a sitcom, the show was, in fact, a prediction of how future humans would live, of what technologies would fill their pockets and furnish their homes.
杰森一家准确预测了视频通话、平板电视、手机、3D 打印和智能手表；所有这些技术在 1962 年都令人难以置信，但到 2022 年却无处不在。然而，有一项技术我们完全未能创造，一项尚未实现的未来主义壮举：名为 Rosey 的自主机器人。
The Jetsons correctly predicted video calls, flat-screen TVs, cell phones, 3D printing, and smartwatches; all technologies that were unbelievable in 1962 and yet were ubiquitous by 2022. However, there is one technology that we have entirely failed to create, one futurist feat that has not yet come to fruition: the autonomous robot named Rosey.
Rosey 是 Jetson 一家的看护人，照看孩子并打理家务。当当时只有 6 岁的 Elroy 在学校里苦苦挣扎时，是 Rosey 帮助他完成家庭作业。当他们 15 岁的女儿 Judy 需要帮助学习开车时，是 Rosey 教她开车。Rosey 做饭、摆桌子、洗碗。Rosey 忠诚、敏感，而且善于开玩笑。她能发现家庭中酝酿的争吵和误解，并介入，帮助人们了解彼此的观点。有一次，Elroy 为他的母亲写的一首诗让她感动得流下了眼泪。在一集中，Rosey 自己甚至坠入了爱河。
Rosey was a caretaker for the Jetson family, watching after the children and tending to the home. When Elroy—then six years old—was struggling in school, it was Rosey who helped him with his homework. When their fifteen-year-old daughter, Judy, needed help learning how to drive, it was Rosey who gave her lessons. Rosey cooked meals, set the table, and did the dishes. Rosey was loyal, sensitive, and quick with a joke. She identified brewing family tiffs and misunderstandings, intervening to help individuals see one another’s perspective. At one time, she was moved to tears by a poem Elroy wrote for his mother. Rosey herself, in one episode, even fell in love.
换句话说，Rosey 拥有人类的智慧。不仅仅是推理、常识和执行复杂任务所需的运动技能不仅要具备在现实世界中完成任务的能力，还要具备成功驾驭社交世界所需的同理心、换位思考能力和社交技巧。用简·杰森的话来说，罗西“就像家人一样。”
In other words, Rosey had the intelligence of a human. Not just the reasoning, common sense, and motor skills needed to perform complex tasks in the physical world, but also the empathizing, perspective taking, and social finesse needed to successfully navigate our social world. In the words of Jane Jetson, Rosey was “just like one of the family.”
尽管《杰森一家》正确地预测了手机和智能手表的出现，但我们仍然没有像 Rosey 这样的东西。截至本书付印时，Rosey 甚至最基本的行为仍然遥不可及。众所周知，第一家制造出可以简单地将碗碟装进洗碗机的机器人的公司将立即拥有畅销产品。所有试图做到这一点的尝试都失败了。这从根本上来说不是机械问题；而是一个智力问题——识别水槽中的物体、正确拾起它们并在不损坏任何东西的情况下将它们装进洗碗机的能力已被证明比以前想象的要困难得多。
Although the The Jetsons correctly predicted cell phones and smartwatches, we still don’t have anything like Rosey. As of this book going to print, even Rosey’s most basic behaviors are still out of reach. It is no secret that the first company to build a robot that can simply load a dishwasher will immediately have a bestselling product. All attempts to do this have failed. It isn’t fundamentally a mechanical problem; it’s an intellectual one—the ability to identify objects in a sink, pick them up appropriately, and load them without breaking anything has proven far more difficult than previously thought.
当然，尽管我们还没有 Rosey，但自 1962 年以来，人工智能 (AI) 领域的进步令人瞩目。人工智能现在可以在包括国际象棋和围棋在内的众多技巧游戏中击败世界上最优秀的人类。人工智能可以像人类放射科医生一样识别放射图像中的肿瘤。人工智能即将实现自动驾驶汽车。而且，在过去几年中，大型语言模型的新进展使 2022 年秋季推出的 ChatGPT 等产品能够创作诗歌、随意在语言之间翻译，甚至编写代码。令地球上的每一位高中老师感到懊恼的是，ChatGPT 可以立即就几乎任何勇敢的学生可能问到的话题撰写一篇写得非常好且原创的文章。ChatGPT 甚至可以通过律师资格考试，得分超过 90% 的律师。
Of course, even though we do not yet have Rosey, the progress in the field of artificial intelligence (AI) since 1962 has been remarkable. AI can now beat the best humans in the world at numerous games of skill, including chess and Go. AI can recognize tumors in radiology images as well as human radiologists. AI is on the cusp of autonomously driving cars. And as of the last few years, new advancements in large language models are enabling products like ChatGPT, which launched in fall 2022, to compose poetry, translate between languages at will, and even write code. To the chagrin of every high school teacher on planet Earth, ChatGPT can instantly compose a remarkably well written and original essay on almost any topic that an intrepid student might ask of it. ChatGPT can even pass the bar exam, scoring better than 90 percent of lawyers.
纵观人工智能的漫长历程，我们始终难以判断我们距离创造出人类水平的智能还有多远。在 20 世纪 60 年代问题解决算法取得早期成功之后，人工智能先驱马文·明斯基 (Marvin Minsky) 曾宣称：“3 到 8 年后，我们将拥有一台拥有普通人类一般智能的机器。”但这并没有发生。在 20 世纪 80 年代专家系统取得成功之后，《商业周刊》宣称“人工智能：它来了”。此后不久，进展停滞不前。而现在，随着大型语言模型的进步，许多研究人员再次宣称“游戏结束了”，因为我们“即将实现人类级别的人工智能。”那么，我们是否终于即将创造出像 Rosey 一样的人类人工智能，还是大型语言像 ChatGPT 这样的模型只是未来几十年漫长旅程中取得的最新成就吗？
Across this long arrow of AI achievements, it has always been hard to tell how close we are to creating human-level intelligence. After the early successes of problem solving algorithms in the 1960s, the AI pioneer Marvin Minsky famously proclaimed that “from three to eight years we will have a machine with the general intelligence of an average human being.” It did not happen. After the successes of expert systems in the 1980s, BusinessWeek proclaimed “AI: it’s here.” Progress stalled shortly thereafter. And now with advancements in large language models, many researchers have again proclaimed that the “game is over” because we are “on the verge of achieving human-level AI.” Which is it: Are we finally on the cusp of creating human-like artificial intelligence like Rosey, or are large language models like ChatGPT just the most recent achievement in a long journey that will stretch on for decades to come?
在此过程中，随着人工智能变得越来越聪明，衡量我们朝着这一目标前进的进展也变得越来越困难。如果人工智能系统在某项任务上的表现优于人类，是否意味着人工智能系统已经掌握了人类解决任务的方法？计算器——能够比人类更快地处理数字——真的懂数学吗？ChatGPT——在律师资格考试中的得分比大多数律师都高——真的懂法律吗？我们如何分辨出差异？在什么情况下，如果有的话，这种差异甚至重要？
Along this journey, as AI keeps getting smarter, it is becoming harder to measure our progress toward this goal. If an AI system outperforms humans on a task, does it mean that the AI system has captured how humans solve the task? Does a calculator—capable of crunching numbers faster than a human—actually understand math? Does ChatGPT—scoring better on the bar exam than most lawyers—actually understand the law? How can we tell the difference, and in what circumstances, if any, does the difference even matter?
2021 年，在 ChatGPT（一款如今正在社会各个角落迅速普及的聊天机器人）发布的一年多前，我正在使用它的前身，一个名为 GPT-3 的大型语言模型。GPT-3 接受了大量文本（大到整个互联网）的训练，然后使用这个语料库尝试对提示进行模式匹配，找出最有可能的响应。当被问到“狗心情不好的两个原因是什么？”时，它回答说：“狗心情不好的两个原因是饿了或热了。”这些系统的新架构使它们能够以至少看起来相当高的智能程度回答问题。这些模型能够将它们读到的事实（例如关于狗的维基百科页面和其他关于心情不好原因的页面）推广到它们从未见过的新问题上。 2021 年，我正在探索这些新语言模型的可能应用——它们能否用于提供新的心理健康支持系统、更无缝的客户服务或更民主化的医疗信息获取方式？
In 2021, over a year before the release of ChatGPT—the chatbot that is now rapidly proliferating throughout every nook and cranny of society—I was using its precursor, a large language model called GPT-3. GPT-3 was trained on large quantities of text (large as in the entire internet), and then used this corpus to try to pattern match the most likely response to a prompt. When asked, “What are two reasons that a dog might be in a bad mood?” it responded, “Two reasons a dog might be in a bad mood are if it is hungry or if it is hot.” Something about the new architecture of these systems enabled them to answer questions with what at least seemed like a remarkable degree of intelligence. These models were able to generalize facts they had read about (like the Wikipedia pages about dogs and other pages about causes of bad moods) to new questions they had never seen. In 2021, I was exploring possible applications of these new language models—could they be used to provide new support systems for mental health, or more seamless customer service, or more democratized access to medical information?
我与 GPT-3 的互动越多，我就越着迷于它的成功和错误。在某些方面，它很聪明，在其他方面，它却愚蠢得离谱。让 GPT-3 写一篇关于 18 世纪马铃薯种植及其与全球化的关系的文章，你会得到一篇出人意料的连贯文章。问它一个关于人们在地下室里可能会看到什么的常识性问题，它会给出毫无意义的回答。*为什么可以GPT-3 能正确回答某些问题，而不能回答其他问题？它捕捉到了人类智能的哪些特征，又遗漏了哪些特征？为什么随着人工智能发展的不断加速，一些在一年内难以回答的问题在随后的几年里变得容易了？事实上，截至本书付印之时，GPT-3 的新升级版 GPT-4 已于 2023 年初发布，可以正确回答许多困扰 GPT-3 的问题。然而，正如我们将在本书中看到的那样，GPT-4 仍未能捕捉到人类智能的基本特征——关于人类大脑中正在发生的事情。
The more I interacted with GPT-3, the more mesmerized I became by both its successes and mistakes. In some ways it was brilliant, in other ways it was oddly dumb. Ask GPT-3 to write an essay about eighteenth-century potato farming and its relationship to globalization, and you will get a surprisingly coherent essay. Ask it a commonsense question about what someone might see in a basement, and it answers nonsensically.* Why could GPT-3 correctly answer some questions and not others? What features of human intelligence does it capture, and which is it missing? And why, as AI development continues to accelerate, are some questions that were hard to answer in one year becoming easy in subsequent years? Indeed, as of this book going to print, the new and upgraded version of GPT-3, called GPT-4, released in early 2023, can correctly answer many questions that beguiled GPT-3. And yet still, as we will see in this book, GPT-4 fails to capture essential features of human intelligence—about something going on in the human brain.
确实，人工智能与人类智能之间的差异令人困惑不已。为什么人工智能可以在国际象棋比赛中击败地球上的任何人，但在装载洗碗机方面却不如六岁儿童？
Indeed, the discrepancies between artificial intelligence and human intelligence are nothing short of perplexing. Why is it that AI can crush any human on earth in a game of chess but can’t load a dishwasher better than a six-year-old?
我们很难回答这些问题，因为我们还不了解我们试图重建的东西。从本质上讲，所有这些问题都不是有关人工智能的问题，而是有关人类智能本身的本质的问题——它是如何运作的，为什么它会这样运作，以及我们很快就会看到的，最重要的是，它是如何产生的。
We struggle to answer these questions because we don’t yet understand the thing we are trying to re-create. All of these questions are, in essence, not questions about AI, but about the nature of human intelligence itself—how it works, why it works the way it does, and as we will soon see, most importantly, how it came to be.
自然的暗示
Nature’s Hint
当人类想要了解飞行时，我们最初的灵感来自鸟类；当乔治·德·梅斯特拉尔发明维可牢尼龙搭扣时，他的灵感来自牛蒡果实；当本杰明·富兰克林试图探索电时，他最初的理解火花来自闪电。在人类创新的整个历史中，大自然一直是一位奇妙的向导。
When humanity wanted to understand flight, we garnered our first inspiration from birds; when George de Mestral invented Velcro, he got the idea from burdock fruits; when Benjamin Franklin sought to explore electricity, his first sparks of understanding came from lightning. Nature has, throughout the history of human innovation, long been a wondrous guide.
大自然也为我们提供了有关智能如何运作的线索——其中最清晰的当然是人类大脑。但在这方面，人工智能与其他技术创新不同；事实证明，大脑比人类大脑更大脑的运作方式比翅膀和闪电更笨重、更难破解。几千年来，科学家们一直在研究大脑的运作方式，虽然我们取得了一些进展，但还没有找到令人满意的答案。
Nature also offers us clues as to how intelligence works—the clearest locus of which is, of course, the human brain. But in this way, AI is unlike these other technological innovations; the brain has proven to be more unwieldy and harder to decipher than either wings or lightning. Scientists have been investigating how the brain works for millennia, and while we have made progress, we do not yet have satisfying answers.
问题在于复杂性。
The problem is complexity.
人类大脑包含 860 亿个神经元和超过 100 万亿个连接。每个连接都非常小——宽度不到 30 纳米——即使在最强大的显微镜下也几乎看不到。这些连接杂乱无章地聚集在一起——在一立方毫米（一美分硬币上一个字母的宽度）内，就有超过十亿个连接。
The human brain contains eighty-six billion neurons and over a hundred trillion connections. Each of those connections is so minuscule—less than thirty nanometers wide—that they can barely be seen under even the most powerful microscopes. These connections are bunched together in a tangled mess—within a single cubic millimeter (the width of a single letter on a penny), there are over one billion connections.
但神经元连接的数量仅仅是大脑复杂性的一个方面。即使我们绘制出每个神经元的连接图，我们也远远不能理解大脑是如何工作的。与计算机中的电子连接不同，在每一个神经连接中，所有电线都使用相同的信号（电子）进行通信。大脑中会传递数百种不同的化学物质，每种都有完全不同的效果。两个神经元相互连接这一简单事实并不能告诉我们它们在传递什么。最糟糕的是，这些连接本身也处于不断变化的状态，一些神经元会分支出来形成新的连接，而另一些则会收缩和移除旧的连接。总之，这使得逆向工程大脑的工作方式成为一项艰巨的任务。
But the sheer number of connections is only one aspect of what makes the brain complex; even if we mapped the wiring of each neuron we would still be far from understanding how the brain works. Unlike the electrical connections in your computer, where wires all communicate using the same signal—electrons—across each of these neural connections, hundreds of different chemicals are passed, each with completely different effects. The simple fact that two neurons connect to each other tells us little about what they are communicating. And worst of all, these connections themselves are in a constant state of change, with some neurons branching out and forming new connections, while others are retracting and removing old ones. Altogether, this makes reverse engineering how the brain works an ungodly task.
研究大脑既诱人又令人恼火。眼睛后面一英寸的地方是宇宙中最令人惊叹的奇迹。它蕴藏着关于智能本质、构建类似人类的人工智能、人类为什么会这样思考和行为的秘密。它就在那里，每年随着每个新生儿的出生，大脑都会被重建数百万次。我们可以触摸它、握住它、解剖它，我们确实是由它组成的，但它的秘密仍然遥不可及，隐藏在显而易见的地方。
Studying the brain is both tantalizing and infuriating. One inch behind your eyes is the most awe-inspiring marvel of the universe. It houses the secrets to the nature of intelligence, to building humanlike artificial intelligence, to why we humans think and behave the way we do. It is right there, reconstructed millions of times per year with every newly born human. We can touch it, hold it, dissect it, we are literally made of it, and yet its secrets remain out of reach, hidden in plain sight.
如果我们想逆向工程大脑的工作方式，如果我们想建造 Rosey，如果我们想揭示人类智能的隐藏本质，也许人类大脑并不是大自然的最佳线索。虽然了解人类大脑最直观的地方自然是人类大脑本身，但与直觉相反，这可能是最后一个要研究的地方。最好的起点可能是深藏在尘土中的化石地壳、隐藏在整个动物界细胞内的微观基因以及地球上许多其他动物的大脑中。
If we want to reverse-engineer how the brain works, if we want to build Rosey, if we want to uncover the hidden nature of human intelligence, perhaps the human brain is not nature’s best clue. While the most intuitive place to look to understand the human brain is, naturally, inside the human brain itself, counterintuitively, this may be the last place to look. The best place to start may be in dusty fossils deep in the Earth’s crust, in microscopic genes tucked away inside cells throughout the animal kingdom, and in the brains of the many other animals that populate our planet.
换句话说，答案可能不在现在，而是在很久以前隐藏的残余中。
In other words, the answer might not be in the present, but in the hidden remnants of a long time past.
失踪的大脑博物馆
The Missing Museum of Brains
我一直坚信，让人工智能发挥作用的唯一方法是以类似进入人类大脑。
I have always been convinced that the only way to get artificial intelligence to work is to do the computation in a way similar to the human brain.
—杰弗里·辛顿（多伦多大学教授，被认为是“人工智能教父”之一）
—GEOFFREY HINTON (PROFESSOR AT UNIVERSITY OF TORONTO, CONSIDERED ONE OF THE “GODFATHERS OF AI”)
人类可以驾驶宇宙飞船、分裂原子、编辑基因。其他动物甚至还没有发明轮子。
Humans fly spaceships, split atoms, and edit genes. No other animal has even invented the wheel.
由于人类拥有丰富的发明创造，你可能会认为我们几乎无法从其他动物的大脑中学到什么。你可能会认为人类的大脑是独一无二的，与其他动物的大脑完全不同，某种特殊的大脑结构可能是我们聪明的秘诀。但事实并非如此。
Because of humanity’s larger résumé of inventions, you might think that we would have little to learn from the brains of other animals. You might think that the human brain would be entirely unique and nothing like the brains of other animals, that some special brain structure would be the secret to our cleverness. But this is not what we see.
当我们研究其他动物的大脑时，最令人震惊的是它们的大脑与我们的大脑有多么相似。除了大小之外，我们的大脑和黑猩猩的大脑几乎没有什么区别。我们的大脑和老鼠的大脑之间的区别只是少数的大脑变化。鱼的大脑几乎与我们的大脑具有相同的结构。
What is most striking when we examine the brains of other animals is how remarkably similar their brains are to our own. The difference between our brain and a chimpanzee’s brain, besides size, is barely anything. The difference between our brain and a rat’s brain is only a handful of brain modifications. The brain of a fish has almost all the same structures as our brain.
动物界中大脑的这些相似性意味着一些重要的事情。它们是线索。关于智力本质的线索。关于我们自己的线索。关于我们过去的线索。
These similarities in brains across the animal kingdom mean something important. They are clues. Clues about the nature of intelligence. Clues about ourselves. Clues about our past.
尽管如今的大脑很复杂，但大脑并非一直如此。大脑诞生于无意识的混乱进化过程；特征中的微小随机变异被选择或被剔除，取决于它们是否支持生命形式的进一步繁殖。
Although today brains are complex, they were not always so. The brain emerged from the unthinking chaotic process of evolution; small random variations in traits were selected for or pruned away depending on whether they supported the further reproduction of the life-form.
在进化过程中，系统从简单开始，复杂性逐渐显现时间。*第一个大脑——动物头部的第一个神经元集合——出现在六亿年前的一条米粒大小的蠕虫身上。这种蠕虫是所有现代有脑动物的祖先。经过数亿年的进化，通过数万亿次细微的线路调整，她简单的大脑变成了现代大脑的多样化组合。这种古老蠕虫后代的一个谱系导致了我们头脑中的大脑。
In evolution, systems start simple, and complexity emerges only over time.* The first brain—the first collection of neurons in the head of an animal—appeared six hundred million years ago in a worm the size of a grain of rice. This worm was the ancestor of all modern brain-endowed animals. Over hundreds of millions of years of evolutionary tinkering, through trillions of small tweaks in wiring, her simple brain was transformed into the diverse portfolio of modern brains. One lineage of this ancient worm’s descendants led to the brain in our heads.
如果我们能回到过去，研究这个最初的大脑，了解它是如何工作的，以及它能带来哪些技巧，那就好了。如果我们能追溯到人类大脑的谱系中复杂化的演变，观察发生的每一个物理变化以及它所提供的智力能力，那就好了。如果我们能做到这一点，我们也许能够掌握最终出现的复杂性。事实上，正如生物学家 Theodosius Dobzhansky 的名言：“生物学中的一切都没有意义，除非从进化的角度看。”
If only we could go back in time and examine this first brain to understand how it worked and what tricks it enabled. If only we could then track the complexification forward in the lineage that led to the human brain, observing each physical modification that occurred and the intellectual abilities it afforded. If we could do this, we might be able to grasp the complexity that eventually emerged. Indeed, as the biologist Theodosius Dobzhansky famously said, “Nothing in biology makes sense except in the light of evolution.”
甚至达尔文也幻想过重建这样一个故事。他在《物种起源》的结尾幻想着未来“心理学将建立在一个新的基础之上，即通过渐进的方式获得每一种心理力量和能力。”达尔文逝世一百五十年后，这也许最终成为可能。
Even Darwin fantasized about reconstructing such a story. He ends his Origin of Species fantasizing about a future when “psychology will be based on a new foundation, that of the necessary acquirement of each mental power and capacity by gradation.” One hundred fifty years after Darwin, this may finally be possible.
虽然我们没有时光机，但原则上我们可以进行时间旅行。仅在过去十年中，进化神经科学家就在重建我们祖先的大脑方面取得了令人难以置信的进展。他们这样做的方法之一是通过化石记录——科学家可以使用古代生物的化石头骨来逆向工程其大脑结构。重建我们祖先大脑的另一种方法是检查动物界中其他动物的大脑。
Although we have no time machines, we can, in principle, engage in time travel. In just the past decade, evolutionary neuroscientists have made incredible progress in reconstructing the brains of our ancestors. One way they do this is through the fossil record—scientists can use the fossilized skulls of ancient creatures to reverse-engineer the structure of their brains. Another way to reconstruct the brains of our ancestors is by examining the brains of other animals in the animal kingdom.
动物界的大脑之所以如此相似，是因为它们都来自共同祖先的共同根源。动物界的每一个大脑都是我们祖先大脑的一点线索；每个大脑不仅是一台机器，还是一个时间胶囊，里面装满了之前数万亿个思想的隐藏线索。通过检查通过了解这些其他动物所具有的智力特长和不具有的智力特长，我们不仅可以开始重建我们祖先的大脑，还可以确定这些古老的大脑赋予了它们什么样的智力。我们可以一起开始追踪每种智力能力的获得过程。
The reason why brains across the animal kingdom are so similar is that they all derive from common roots in shared ancestors. Every brain in the animal kingdom is a little clue as to what the brains of our ancestors looked like; each brain is not only a machine but a time capsule filled with hidden hints of the trillions of minds that came before. And by examining the intellectual feats these other animals share and those they do not, we can begin to not only reconstruct the brains of our ancestors, but also determine what intellectual abilities these ancient brains afforded them. Together, we can begin to trace acquirement of each mental power by gradation.
当然，这一切仍处于进行中，但故事正变得越来越清晰。
It is all, of course, still a work in progress, but the story is becoming tantalizingly clear.
层次的神话
The Myth of Layers
我并不是第一个提出用进化框架来理解人类大脑的人。这种框架由来已久。最著名的框架是由神经科学家保罗·麦克莱恩在 20 世纪 60 年代提出的。麦克莱恩假设人类大脑由三层组成（因此是三位一体），每一层都建立在另一层之上：新皮质是最近进化的，其上是边缘系统，其上是进化较早的，其下是爬行动物的大脑，其上是最早进化的。
I am hardly the first to propose an evolutionary framework for understanding the human brain. There is a long tradition of such frameworks. The most famous was formulated in the 1960s by the neuroscientist Paul MacLean. MacLean hypothesized that the human brain was made of three layers (hence triune), each built on top of another: the neocortex, which evolved most recently, on top of the limbic system, which evolved earlier, on top of the reptile brain, which evolved first.
麦克莱恩认为，爬行动物的大脑是我们基本生存本能（如攻击性和领土意识）的中心。边缘系统被认为是情绪的中心，如恐惧、父母依恋、性欲和饥饿。大脑皮层被认为是认知的中心，赋予我们语言、抽象、计划和感知。麦克莱恩的框架表明，爬行动物只有爬行动物的大脑，老鼠和兔子等哺乳动物有爬行动物的大脑和边缘系统，而我们人类则拥有这三个系统。事实上，对他来说，这“三种进化形态可以被想象成三台相互连接的生物计算机，每台都有自己的特殊智能、自己的主观性、自己的时间和空间感以及自己的记忆，电机和其他功能。”
MacLean argued that the reptile brain was the center of our basic survival instincts, such as aggression and territoriality. The limbic system was supposedly the center of emotions, such as fear, parental attachment, sexual desire, and hunger. And the neocortex was supposedly the center of cognition, gifting us with language, abstraction, planning, and perception. MacLean’s framework suggested that reptiles had only a reptile brain, mammals like rats and rabbits had a reptile brain and a limbic system, and we humans had all three systems. Indeed, to him, these “three evolutionary formations might be imagined as three interconnected biological computers, with each having its own special intelligence, its own subjectivity, its own sense of time and space, and its own memory, motor, and other functions.”
问题在于，麦克莱恩的“三位一体大脑假说”在很大程度上已被否定——不是因为它不准确（所有框架都是不准确的），而是因为它导致了关于大脑如何进化的错误结论以及它的工作原理。隐含的大脑解剖结构是错误的；爬行动物的大脑不仅由麦克莱恩所说的“爬行动物大脑”的结构组成；爬行动物还有自己的边缘系统。功能划分被证明是错误的；生存本能、情感和认知并没有清晰的界限——它们来自跨越所有这三个假定层次的不同系统网络。而隐含的进化故事也被证明是错误的。你的脑袋里没有爬行动物的大脑；进化不是简单地将一个系统叠加在另一个系统之上而不对现有系统进行任何修改。
The problem is that MacLean’s Triune Brain Hypothesis has been largely discredited—not because it is inexact (all frameworks are inexact), but because it leads to the wrong conclusions about how the brain evolved and how it works. The implied brain anatomy is wrong; the brains of reptiles are not only made up of the structures MacLean referred to as the “reptile brain”; reptiles also have their own version of a limbic system. The functional divisions proved wrong; survival instincts, emotions, and cognition do not delineate cleanly—they emerge from diverse networks of systems spanning all three of these supposed layers. And the implied evolutionary story turned out to be wrong. You do not have a reptile brain in your head; evolution did not work by simply layering one system on top of another without any modifications to the existing systems.
图 1：麦克莱恩的三位一体大脑
Figure 1: MacLean’s triune brain
马克斯·贝内特 (Max Bennett) 绘制的人物（受到麦克莱恩 (MacLean) 作品中类似人物的启发）
Figure by Max Bennett (inspired by similar figures found in MacLean’s work)
但即使麦克莱恩的三重大脑理论更接近事实，它最大的问题在于，其功能划分对于我们来说并不是特别有用。如果我们的目标是逆向工程人类大脑以了解智力的本质，那么麦克莱恩的三个系统太过宽泛，而归因于它们的功能又太过模糊，甚至无法为我们提供切入点。
But even if MacLean’s triune brain had turned out to be closer to the truth, its biggest problem is that its functional divisions aren’t particularly useful for our purposes. If our goal is to reverse-engineer the human brain to understand the nature of intelligence, MacLean’s three systems are too broad and the functions attributed to them too vague to provide us with even a point at which to start.
我们需要将我们对大脑如何运作及其如何进化的理解建立在我们对智能如何运作的理解之上——为此我们必须关注人工智能领域。人工智能和大脑之间的关系是双向的；虽然大脑肯定可以教会我们很多关于如何创造人工智能的知识，但人工智能也可以教会我们关于大脑的知识。如果我们认为大脑的某个部分使用某种特定的算法，但当我们在机器中实现它时，这种算法不起作用，这就给了我们大脑可能不是这样运作的证据。相反，如果我们发现一种在人工智能系统中运行良好的算法，并且我们发现这些算法的属性之间存在相似之处以及动物大脑的特性，这为我们提供了一些证据，证明大脑确实可能以这种方式运作。
We need to ground our understanding of how the brain works and how it evolved in our understanding of how intelligence works—for which we must look to the field of artificial intelligence. The relationship between AI and the brain goes both ways; while the brain can surely teach us much about how to create artificial humanlike intelligence, AI can also teach us about the brain. If we think some part of the brain uses some specific algorithm but that algorithm doesn’t work when we implement it in machines, this gives us evidence that the brain might not work this way. Conversely, if we find an algorithm that works well in AI systems, and we find parallels between the properties of these algorithms and properties of animal brains, this gives us some evidence that the brain might indeed work this way.
物理学家理查德·费曼在去世前不久在黑板上留下了这样的话：“我不能创造的东西，我就不理解。”大脑是我们如何构建人工智能的指导灵感，而人工智能则是我们对大脑了解程度的试金石。
The physicist Richard Feynman left the following on a blackboard shortly before his death: “What I cannot create, I do not understand.” The brain is our guiding inspiration for how to build AI, and AI is our litmus test for how well we understand the brain.
我们需要一个新的大脑进化故事，它不仅要建立在对大脑解剖结构随时间如何变化的现代理解之上，还要建立在对智力本身的现代理解之上。
We need a new evolutionary story of the brain, one grounded not only in a modern understanding of how brain anatomy changed over time, but also in a modern understanding of intelligence itself.
五大突破
The Five Breakthroughs
让我们从人工智能大鼠级智能（ARI）开始，然后转向人工智能猫级智能（ACI），然后直至达到人类级别的人工智能（AHI）。
Let’s start with Artificial Rat–level intelligence (ARI), then move on to Artificial Cat–level intelligence (ACI), and so on to Artificial Human–level Intelligence (AHI).
— META 人工智能负责人YANN LECUN
—YANN LECUN, HEAD OF AI AT META
我们要讲述很长的进化史——四十亿年。我们不会记录每一个微小的调整，而是记录重大的进化突破。事实上，作为初步的近似——这个故事的第一个模板——人类大脑的整个进化过程可以合理地概括为仅仅五个突破的顶峰，从最初的大脑开始一直到人类大脑。
We have a lot of evolutionary history to cover—four billion years. Instead of chronicling each minor adjustment, we will be chronicling the major evolutionary breakthroughs. In fact, as an initial approximation—a first template of this story—the entirety of the human brain’s evolution can be reasonably summarized as the culmination of only five breakthroughs, starting from the very first brains and going all the way to human brains.
这五项突破是本书的组织图，它们构成了我们穿越时空的冒险旅程。每项突破都源自新的大脑改造，并为动物提供了一系列新的智力能力。本书分为五个部分，每项突破各占一部分。在每一部分中，我都会描述这些能力为何进化、如何发挥作用以及它们在当今人类大脑中如何体现。
These five breakthroughs are the organizing map to our book, and they make up our itinerary for our adventure back in time. Each breakthrough emerged from new sets of brain modifications and equipped animals with a new portfolio of intellectual abilities. This book is divided into five parts, one for each breakthrough. In each section, I will describe why these abilities evolved, how they worked, and how they still manifest in human brains today.
随后的每一次突破都是建立在前人的基础上，并为后来的突破奠定了基础。过去的创新促成了未来的创新。正是通过这一系列有序的修改，大脑的进化故事帮助我们理解最终出现的复杂性。
Each subsequent breakthrough was built on the foundation of those that came before and provided the foundation for those that would follow. Past innovations enabled future innovations. It is through this ordered set of modifications that the evolutionary story of the brain helps us make sense of the complexity that eventually emerged.
但仅从我们祖先大脑的生物学角度看，这个故事是无法如实地重述的。这些突破总是出现在我们的祖先面临极端情况或陷入强大反馈循环的时期。正是这些压力导致了大脑的快速重构。如果不了解我们祖先的考验和胜利，我们就无法理解大脑进化的突破：他们战胜的掠食者、他们忍受的环境灾难以及他们为生存而求助的绝望环境。
But this story cannot be faithfully retold by considering only the biology of our ancestors’ brains. These breakthroughs always emerged from periods when our ancestors faced extreme situations or got caught in powerful feedback loops. It was these pressures that led to rapid reconfigurations of brains. We cannot understand the breakthroughs in brain evolution without also understanding the trials and triumphs of our ancestors: the predators they outwitted, the environmental calamities they endured, and the desperate niches they turned to for survival.
至关重要的是，我们将以人工智能领域目前已知的知识为基础实现这些突破，因为生物智能领域的许多突破与我们在人工智能领域学到的知识相似。其中一些突破代表了我们在人工智能领域熟知的智力技巧，而其他技巧仍然超出我们的理解范围。通过这种方式，大脑的进化故事或许可以揭示我们在发展类似人类的人工智能时可能错过的突破。也许它将揭示一些大自然隐藏的线索。
And crucially, we will ground these breakthroughs in what is currently known in the field of AI, for many of these breakthroughs in biological intelligence have parallels to what we have learned in artificial intelligence. Some of these breakthroughs represent intellectual tricks we understand well in AI, while other tricks still lay beyond our understanding. And in this way, perhaps the evolutionary story of the brain can shed light on what breakthroughs we may have missed in the development of artificial humanlike intelligence. Perhaps it will reveal some of nature’s hidden clues.
我
Me
我希望我能告诉你，我写这本书是因为我一生都在思考大脑的进化，并试图制造智能机器人。但我不是神经科学家，也不是机器人专家，甚至不是科学家。我写这本书是因为我想读这本书。
I wish I could tell you that I wrote this book because I have spent my whole life pondering the evolution of the brain and trying to build intelligent robots. But I am not a neuroscientist or a roboticist or even a scientist. I wrote this book because I wanted to read this book.
我在尝试将人工智能系统应用于现实世界的问题时，发现了人类和人工智能之间令人困惑的差异。我的大部分职业生涯都在我共同创办的 Bluecore 公司度过；我们开发软件和人工智能系统，帮助世界上一些最大的品牌个性化营销。我们的软件帮助预测消费者在知道自己想要什么之前会买什么。我们只是无数开始使用人工智能系统新进展的公司中的一小部分。但所有这些项目，无论大小，都是由同样令人困惑的问题所塑造的。
I came to the perplexing discrepancy between human and artificial intelligence by trying to apply AI systems to real-world problems. I spent the bulk of my career at a company I cofounded named Bluecore; we built software and AI systems to help some of the largest brands in the world personalize their marketing. Our software helped predict what consumers would buy before they knew what they wanted. We were merely one tiny part in a sea of countless companies beginning to use the new advances in AI systems. But all these many projects, both big and small, were shaped by the same perplexing questions.
在商业化 AI 系统时，业务团队和机器学习团队之间最终会进行一系列会议。业务团队团队寻找新人工智能系统有价值的应用，而只有机器学习团队明白哪些应用是可行的。这些会议往往会揭示我们对自己对智能的理解程度的错误直觉。商界人士探究在他们看来简单易懂的人工智能系统应用。但通常，这些任务看起来很简单只是因为它们对我们的大脑来说很简单。机器学习人员然后耐心地向业务团队解释为什么看似简单的想法实际上难度极大。这些争论在每个新项目中都会反复进行。正是从这些对现代人工智能系统能够延伸到多远以及它们的不足之处的探索中，我产生了对大脑最初的好奇心。
When commercializing AI systems, there is eventually a series of meetings between business teams and machine learning teams. The business teams look for applications of new AI systems that would be valuable, while only the machine learning teams understand what applications would be feasible. These meetings often reveal our mistaken intuitions about how much we understand about intelligence. Businesspeople probe for applications of AI systems that seem straightforward to them. But frequently, these tasks seem straightforward only because they are straightforward for our brains. Machine learning people then patiently explain to the business team why the idea that seems simple is, in fact, astronomically difficult. And these debates go back and forth with every new project. It was from these explorations into how far we could stretch modern AI systems and the surprising places where they fall short that I developed my original curiosity about the brain.
当然，我也是人类，和你一样，我也有人类的大脑。所以我很容易对这个决定人类体验的器官着迷。大脑不仅能解答智力的本质，还能解答我们为什么会这样做。为什么我们经常做出不合理和自欺欺人的选择？为什么我们这个物种有如此悠久的历史，既鼓舞人心的无私，又深不可测的残忍？
Of course, I am also a human and I, like you, have a human brain. So it was easy for me to become fascinated with the organ that defines so much of the human experience. The brain offers answers not only about the nature of intelligence, but also why we behave the way we do. Why do we frequently make irrational and self-defeating choices? Why does our species have such a long recurring history of both inspiring selflessness and unfathomable cruelty?
我的个人项目最初只是试图读书来回答我自己的问题。这最终升级为与神经科学家进行冗长的电子邮件通信，他们慷慨地满足了局外人的好奇心。这项研究和这些通信最终促使我发表了几篇研究论文，最终我决定抽出时间将这些酝酿的想法写成一本书。
My personal project began with merely trying to read books to answer my own questions. This eventually escalated to lengthy email correspondences with neuroscientists who were generous enough to indulge the curiosities of an outsider. This research and these correspondences eventually led me to publish several research papers, which all culminated in the decision to take time off work to turn these brewing ideas into a book.
在整个过程中，我研究得越深入，就越相信可以做出有价值的综合贡献，它可以提供对大脑如何工作、为什么这样工作以及它与现代人工智能系统有何重叠和不同之处的易于理解的介绍；它可以将神经科学和人工智能领域的各种想法整合到一个故事的保护伞下。
Throughout this process, the deeper I went, the more I became convinced that there was a worthwhile synthesis to be contributed, one that could provide an accessible introduction to how the brain works, why it works the way it does, and how it overlaps and differs from modern AI systems; one that could bring various ideas across neuroscience and AI together under an umbrella of a single story.
《智力简史》综合了许多其他人的研究成果。从本质上讲，它只是试图将已经存在的碎片拼凑在一起。我在整本书中尽力给予应有的赞誉，始终致力于赞扬那些进行实际研究的科学家。任何未能做到这点的地方都是无意的。诚然，我忍不住要加入一些我自己的猜测，但当我涉足这一领域时，我会尽量说清楚。
A Brief History of Intelligence is a synthesis of the work of many others. At its heart, it is merely an attempt to put together the pieces that were already there. I have done my best to give due credit throughout the book, always aiming to celebrate those scientists who did the actual research. Anywhere I have failed to do so is unintentional. Admittedly, I couldn’t resist sprinkling in a few speculations of my own, but I will aim to be clear when I step into such territory.
这本书的起源就像大脑本身的起源一样，并非源自事先的规划，而是源自错误的开始和错误转折的混乱过程，源自机会、迭代和幸运的情况。
It is perhaps fitting that the origin of this book, like the origin of the brain itself, came not from prior planning but from a chaotic process of false starts and wrong turns, from chance, iteration, and lucky circumstance.
最后一点（关于梯子和沙文主义）
A Final Point (About Ladders and Chauvinism)
在我们开始回到过去之前，我还有最后一点要说。整个故事中存在着一种危险的误解。
I have one final point to make before we begin our journey back in time. There is a misinterpretation that will loom dangerously between the lines of this entire story.
本书将人类的能力与当今其他动物的能力进行多次比较，但总是会选择那些被认为与我们的祖先最相似的动物。整本书——五大突破框架本身——都是关于人类谱系的故事，关于我们大脑如何形成的故事；人们可以很容易地构建一个关于章鱼或蜜蜂大脑如何形成的故事，它会有自己的曲折和突破。
This book will draw many comparisons between the abilities of humans and those of other animals alive today, but this is always done by picking specifically those animals that are believed to be most similar to our ancestors. This entire book—the five-breakthroughs framework itself—is solely the story of the human lineage, the story of how our brains came to be; one could just as easily construct a story of how the octopus or honeybee brain came to be, and it would have its own twists and turns and its own breakthroughs.
仅仅因为我们的大脑比我们祖先的大脑拥有更多的智力，并不意味着现代人类的大脑在智力上严格优于其他现代动物的大脑。
Just because our brains wield more intellectual abilities than those of our ancestors does not mean that the modern human brain is strictly intellectually superior to those of other modern animals.
进化总是独立地趋向于共同的解决方案。翅膀的创新在昆虫、蝙蝠和鸟类中独立进化；这些生物的共同祖先没有翅膀。人们还认为眼睛已经独立进化了很多次。因此，当我论证一种智力能力（如情景记忆）是在早期哺乳动物中进化而来时，这并不意味着今天只有哺乳动物才具有情景记忆。就像翅膀和眼睛一样，其他生命谱系可能也独立进化出了情景记忆。事实上，我们将在本书中记录的许多智力能力并不是我们谱系所独有的，而是沿着地球进化树的众多分支独立发芽的。
Evolution independently converges on common solutions all the time. The innovation of wings independently evolved in insects, bats, and birds; the common ancestor of these creatures did not have wings. Eyes are also believed to have independently evolved many times. Thus, when I argue that an intellectual ability, such as episodic memory, evolved in early mammals, this does not mean that today only mammals have episodic memory. Like with wings and eyes, other lineages of life may have independently evolved episodic memory. Indeed, many of the intellectual faculties that we will chronicle in this book are not unique to our lineage, but have independently sprouted along numerous branches of earth’s evolutionary tree.
自亚里士多德时代以来，科学家和哲学家们已经构建了现代生物学家称之为“自然等级”（或者，科学家喜欢使用拉丁语术语scala naturae）。亚里士多德创建了所有生命形式的等级制度，人类优于其他哺乳动物，而哺乳动物又优于爬行动物和鱼类，而爬行动物和鱼类又优于昆虫，昆虫又优于植物。
Since the days of Aristotle, scientists and philosophers have constructed what modern biologists refer to as a “scale of nature” (or, since scientists like using Latin terms, scala naturae). Aristotle created a hierarchy of all life-forms with humans being superior to other mammals, who were in turn superior to reptiles and fish, who were in turn superior to insects, who were in turn superior to plants.
图 2
Figure 2
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
即使在发现进化论之后，自然界的等级观念仍然存在。物种存在等级观念是完全错误的。今天活着的所有物种都是活着的；它们的祖先在过去 35 亿年的进化中幸存下来。因此，从这个意义上讲——进化论唯一关心的意义上——今天活着的所有生命形式都并列第一。
Even after the discovery of evolution, the idea of a scale of nature continues to persist. This idea that there is a hierarchy of species is dead wrong. All species alive today are, well, alive; their ancestors survived the last 3.5 billion years of evolution. And thus, in that sense—the only sense that evolution cares about—all life-forms alive today are tied for first place.
物种分为不同的生存利基，每个利基针对不同的事物进行优化。许多利基——事实上，大多数利基——更适合更小、更简单的大脑（或根本没有大脑）。大脑较大的猿类是与蠕虫、细菌或蝴蝶不同的生存策略的结果。但没有一个是“更好的”。从进化的角度来看，等级制度只有两个梯级：一个梯级上有幸存下来的物种，另一个梯级上有未能幸存的物种。
Species fall into different survival niches, each of which optimizes for different things. Many niches—in fact, most niches—are better served by smaller and simpler brains (or no brains at all). Big-brained apes are the result of a different survival strategy than that of worms, bacteria, or butterflies. But none are “better.” In the eyes of evolution, the hierarchy has only two rungs: on one, there are those that survived, and on the other, those that did not.
也许，人们更愿意用智力的某些特定特征来定义。但在这里，排名仍然完全取决于我们衡量的具体智力技能。章鱼在每根触手中都有一个独立的大脑，在多任务处理方面可以胜过人类。鸽子、花栗鼠、金枪鱼，甚至鬣蜥都可以处理视觉信息比人类更快。鱼具有令人难以置信的精确实时处理能力；如果你试图抓住一条鱼，你见过它能以多快的速度穿过岩石迷宫吗？如果人类试图如此快速地穿越障碍赛道，他肯定会摔倒。
Perhaps instead, one wants to define better by some specific feature of intelligence. But here still, the ranking will entirely depend on what specific intellectual skill we are measuring. An octopus has an independent brain in each of its tentacles and can blow a human away at multitasking. Pigeons, chipmunks, tuna, and even iguanas can process visual information faster than a human. Fish have incredibly accurate real-time processing; have you ever seen how fast a fish whips through a maze of rocks if you try to grab it? A human would surely crash if he or she tried to move so quickly through an obstacle course.
我的呼吁：在追溯我们的故事时，我们必须避免认为从过去到未来的复杂化表明现代人类绝对优于现代动物。我们必须避免意外构建自然阶梯。当今所有活着的动物都经历了相同时间的进化。
My appeal: As we trace our story, we must avoid thinking that the complexification from past to future suggests that modern humans are strictly superior to modern animals. We must avoid the accidental construction of a scala naturae. All animals alive today have been undergoing evolution for the same amount of time.
然而，人类当然有其独特之处，因为我们是人类，所以我们对了解自己抱有特殊兴趣是有道理的，我们努力制造出像人类一样的人工智能也是有道理的。所以我希望我们能够参与以人类为中心的故事，而不会沦为人类沙文主义。对于与我们共享地球的任何其他动物，从蜜蜂到鹦鹉再到章鱼，都有同样值得讲述的故事。但我们不会在这里讲述这些故事。这本书只讲述了其中一种智能的故事：它讲述了我们的故事。
However, there are, of course, things that make us humans unique, and because we are human, it makes sense that we hold a special interest in understanding ourselves, and it makes sense that we strive to make artificial humanlike intelligences. So I hope we can engage in a human-centered story without devolving into human chauvinism. There is an equally valid story to be told for any other animal, from honeybees to parrots to octopuses, with which we share our planet. But we will not tell these stories here. This book tells the story of only one of these intelligences: it tells the story of us.
1
1
大脑之前的世界
The World Before Brains
在第一个大脑出现之前，地球上的生命已经存在了很长一段时间——我的意思是很长一段时间，超过 30 亿年。当第一个大脑进化时，生命已经经历了无数次充满挑战和变化的进化周期。在地球生命的宏伟弧线中，大脑的故事不会出现在主要章节中，而是出现在尾声中——大脑只出现在生命故事最近的 15% 中。在大脑出现之前，智力也存在了很长一段时间；正如我们将看到的，生命在其故事的早期就开始表现出智力行为。如果不首先回顾智力本身的进化，我们就无法理解大脑为何以及如何进化。
LIFE EXISTED ON Earth for a long time—and I mean a long time, over three billion years—before the first brain made an appearance. By the time the first brains evolved, life had already persevered through countless evolutionary cycles of challenge and change. In the grand arc of life on Earth, the story of brains would not be found in the main chapters but in the epilogue—brains appeared only in the most recent 15 percent of life’s story. Intelligence too existed for a long time before brains; as we will see, life began exhibiting intelligent behavior early in its story. We cannot understand why and how brains evolved without first reviewing the evolution of intelligence itself.
大约四十亿年前，在没有生命的地球的火山海洋深处，恰到好处的分子汤在微小的角落和缝隙中跳跃。一个不起眼的热液喷口。沸水从海底喷涌而出，将自然产生的核苷酸粉碎在一起，并将它们转化为与今天的 DNA 非常相似的长分子链。这些早期的 DNA 类分子寿命很短；构造它们的火山动能也不可避免地将它们撕裂。这是热力学第二定律的结果。这条牢不可破的物理定律表明，熵——系统中无序的量——总是不可避免地增加；宇宙不得不趋向衰变。在无数随机核苷酸链被构建和摧毁之后，一个幸运的序列被偶然发现，它标志着，至少在地球上，第一次真正反抗看似不可避免的熵的冲击。这种新的 DNA 类分子本身并不是活的，但它执行了生命后来出现的最基本过程：它自我复制了。
Around four billion years ago, deep in the volcanic oceans of a lifeless Earth, just the right soup of molecules were bouncing around the microscopic nooks and crannies of an unremarkable hydrothermal vent. As boiling water burst from the seafloor, it smashed naturally occurring nucleotides together and transformed them into long molecular chains that closely resembled today’s DNA. These early DNA-like molecules were short-lived; the same volcanic kinetic energy that constructed them also inevitably ripped them apart. Such is the consequence of the second law of thermodynamics. That unbreakable law of physics which declares that entropy—the amount of disorder in a system—always and unavoidably increases; the universe cannot help but tend toward decay. After countless random nucleotide chains were constructed and destroyed, a lucky sequence was stumbled upon, one that marked, at least on Earth, the first true rebellion against the seemingly inexorable onslaught of entropy. This new DNA-like molecule wasn’t alive per se, but it performed the most fundamental process by which life would later emerge: it duplicated itself.
虽然这些自我复制的 DNA 类分子也屈服于熵的破坏性影响，但它们不必单独生存才能集体生存——只要它们坚持足够长的时间以产生自己的副本，它们就会本质上持续存在。这就是自我复制的天才之处。随着这些第一批自我复制分子的出现，进化过程的原始版本开始了；任何促进更成功复制的新幸运情况当然都会导致更多的复制。
Although these self-replicating DNA-like molecules also succumbed to the destructive effects of entropy, they didn’t have to survive individually to survive collectively—as long as they endured long enough to create their own copies, they would, in essence, persist. This is the genius of self-replication. With these first self-replicating molecules, a primitive version of the process of evolution began; any new lucky circumstances that facilitated more successful duplication would, of course, lead to more duplicates.
随后发生了两次进化转变，最终形成了生命。第一次是保护性脂质气泡捕获了这些 DNA 分子，其机制与肥皂（也是由脂质制成的）在洗手时自然起泡的机制相同。这些充满 DNA 的微小脂质气泡是细胞（生命的基本单位）的最初版本。
There were two subsequent evolutionary transformations that led to life. The first was when protective lipid bubbles entrapped these DNA molecules using the same mechanism by which soap, also made of lipids, naturally bubbles when you wash your hands. These DNA-filled microscopic lipid bubbles were the first versions of cells, the fundamental unit of life.
第二次进化转变发生在一组基于核苷酸的分子——核糖体——开始将特定的 DNA 序列翻译成特定的氨基酸序列，然后折叠成我们称之为蛋白质的特定三维结构时。一旦产生，这些蛋白质就会在细胞内漂浮或嵌入细胞壁，发挥不同的功能。你可能至少听说过，你的 DNA 是由基因组成的。其实，基因只是 DNA 中编码特定单一蛋白质构建的部分。这是蛋白质合成的发明，也是智慧的火花首次出现的地方。
The second evolutionary transformation occurred when a suite of nucleotide-based molecules—ribosomes—began translating specific sequences of DNA into specific sequences of amino acids that were then folded into specific three-dimensional structures we call proteins. Once produced, these proteins float around inside a cell or are embedded in the wall of the cell fulfilling different functions. You have probably, at least in passing, heard that your DNA is made up of genes. Well, a gene is simply the section of DNA that codes for the construction of a specific and singular protein. This was the invention of protein synthesis, and it is here that the first sparks of intelligence made their appearance.
DNA 相对惰性，可以有效地自我复制，但在操控周围微观世界方面却有限。然而，蛋白质则灵活得多，功能强大得多。在许多方面，蛋白质更像机器而不是分子。蛋白质可以构造和折叠成许多形状——运动隧道、闩锁和其他机器人运动部件——从而可以实现无穷无尽的细胞功能，包括“智能”。
DNA is relatively inert, effective for self-duplication but otherwise limited in its ability to manipulate the microscopic world around it. Proteins, however, are far more flexible and powerful. In many ways, proteins are more machine than molecule. Proteins can be constructed and folded into many shapes—sporting tunnels, latches, and other robotic moving parts—and can thereby subserve endless cellular functions, including “intelligence.”
即使是最简单的单细胞生物（如细菌），也有为运动而设计的蛋白质、将细胞能量转化为推进力的机动引擎、使用不亚于马达的复杂机制旋转的螺旋桨现代船舶的构造。细菌还具有感知蛋白质——当它们检测到外部环境的某些特征（如温度、光线或触觉）时，受体会重塑。
Even the simplest single-celled organisms—such as bacteria—have proteins designed for movement, motorized engines that convert cellular energy into propulsion, rotating propellers using a mechanism no less complex than the motor of a modern boat. Bacteria also have proteins designed for perception—receptors that reshape when they detect certain features of the external environment, such as temperature, light, or touch.
有了运动和感知的蛋白质，早期生命就可以监测并响应外界。细菌可以远离降低成功复制概率的环境，例如温度过高或过低的环境，或破坏 DNA 或细胞膜的化学物质。细菌也可以游向适合繁殖的环境。
Armed with proteins for movement and perception, early life could monitor and respond to the outside world. Bacteria can swim away from environments that lower the probability of successful replication, environments that have, for example, temperatures that are too hot or cold or chemicals that are destructive to DNA or cell membranes. Bacteria can also swim toward environments that are amenable to reproduction.
这样，这些古老的细胞确实拥有原始的智力，不是通过神经元来实现，而是通过复杂的化学级联和蛋白质网络来实现。
And in this way, these ancient cells indeed had a primitive version of intelligence, implemented not in neurons but in a complex network of chemical cascades and proteins.
蛋白质合成的发展不仅催生了智慧的种子，还将 DNA 从单纯的物质转变为存储信息的媒介。DNA 不再是生命本身的自我复制物质，而是转变为构成生命物质的信息基础。DNA 正式成为生命的蓝图，核糖体是生命的工厂，蛋白质是生命的产品。
The development of protein synthesis not only begot the seeds of intelligence but also transformed DNA from mere matter to a medium for storing information. Instead of being the self-replicating stuff of life itself, DNA was transformed into the informational foundation from which the stuff of life is constructed. DNA had officially become life’s blueprint, ribosomes its factory, and proteins its product.
有了这些基础，进化过程便全面启动：DNA 的变异导致蛋白质的变异，进而引发对新细胞机制的进化探索，这些机制通过自然选择，根据其是否进一步支持生存而进行修剪和选择。生命故事发展到这一阶段，我们已经完成了科学家称之为“无生源论”的漫长、尚未复制且神秘的过程：非生物物质（abio）转化为生命（genesis）的过程。
With these foundations in place, the process of evolution was initiated in full force: variations in DNA led to variations in proteins, which led to the evolutionary exploration of new cellular machinery, which, through natural selection, were pruned and selected for based on whether they further supported survival. By this point in life’s story, we have concluded the long, yet-to-be-replicated, and mysterious process scientists call abiogenesis: the process by which nonbiological matter (abio) is converted into life (genesis).
地球的改造
The Terraforming of Earth
此后不久，这些细胞进化成了科学家所称的“最后的共同祖先”，即 LUCA。LUCA 是所有生命的无性别祖先；如今活着的每一种真菌、植物、细菌和动物，包括我们，都是 LUCA 的后代。因此，所有生命都具有 LUCA 的核心特征也就不足为奇了：DNA、蛋白质合成、脂质和碳水化合物。
Shortly thereafter, these cells evolved into what scientists call the “last universal common ancestor,” or LUCA. LUCA was the genderless grandparent of all life; every fungus, plant, bacteria, and animal alive today, including us, descend from LUCA. It is no surprise, then, that all life shares the core features of LUCA: DNA, protein synthesis, lipids, and carbohydrates.
LUCA 生活在大约 35 亿年前，很可能类似于现代细菌的简化版本。事实上，在此之后的很长一段时间里，所有生命都是细菌。再过十亿年——经过数万亿次进化迭代——地球海洋中充满了许多不同种类的微生物，每种都有自己的 DNA 和蛋白质组合。这些早期微生物与现代微生物的一个不同之处在于能量生产系统相互影响。生命的故事，从本质上讲，既关乎能量，也关乎熵。
LUCA, living around 3.5 billion years ago, likely resembled a simpler version of a modern bacteria. And indeed, for a long time after this, all life was bacterial. After a further billion years—through trillions upon trillions of evolutionary iterations—Earth’s oceans were brimming with many diverse species of these microbes, each with its own portfolio of DNA and proteins. One way in which these early microbes differed from one another was in their systems of energy production. The story of life, at its core, is as much about energy as it is about entropy.
维持细胞存活的成本很高。DNA 需要不断修复；蛋白质需要不断补充；细胞复制需要重建许多内部结构。氢是一种在热液喷口附近含量丰富的元素，很可能是用于为这些过程提供资金。但这种以氢为基础的能源系统效率低下，生命只能拼命地获取足够的能量才能生存。经过十多亿年的生命，这种能量匮乏终于结束了，当时一种细菌——蓝藻，也被称为蓝藻发现了一种更有利可图的提取和储存能量的机制：光合作用。
Keeping a cell alive is expensive. DNA requires constant repair; proteins require perpetual replenishment; and cellular duplication requires a reconstruction of many inner structures. Hydrogen, an element abundant near hydrothermal vents, was likely the first fuel used to finance these many processes. But this hydrogen-based energy system was inefficient and left life desperately grasping for enough energy to survive. After more than a billion years of life, this energetic poverty came to an end when a single species of bacteria—the cyanobacteria, also called blue-green algae—found a far more profitable mechanism for extracting and storing energy: photosynthesis.
摄影：Willem van Aken，1993 年 3 月 18 日。图片来自 www.scienceimage.csiro.au/image/4203 CC BY 3.0 许可证。
Photograph by Willem van Aken on March 18, 1993. Figure from www.scienceimage.csiro.au/image/4203 CC BY 3.0 license.
这些早期蓝藻中最令人印象深刻的生物系统不是它们的蛋白质工厂或产品，而是它们的光合作用发电厂——将阳光和二氧化碳转化为糖的结构，然后可以储存并转化为细胞能量。光合作用比以前的细胞系统更有效地提取和储存能量。它为蓝藻提供了丰富的燃料，为它们的复制提供资金。大片海洋很快就被绿色黏糊糊的微生物垫覆盖——数十亿个蓝藻群落在阳光下，吸收二氧化碳，并不断繁殖。
The most impressive biological system in these early cyanobacteria was not their protein factories or their products but their photosynthetic power plants—the structures that converted sunlight and carbon dioxide into sugar, which could then be stored and converted into cellular energy. Photosynthesis was more efficient than prior cellular systems for extracting and storing energy. It provided cyanobacteria with an abundance of fuel with which to finance their duplication. Vast regions of the ocean quickly became covered in greenish gooey microbial mats—colonies of billions of cyanobacteria basking in the sun, sucking up carbon dioxide, and endlessly reproducing.
与大多数能源生产过程一样，从燃烧化石燃料到利用核燃料、光合作用光合作用产生的废气是污染性的。光合作用产生的废气不是二氧化碳或核废料，而是氧气。在此之前，地球没有臭氧层。正是蓝藻通过新发现的光合作用，构建了地球富含氧气的大气层，并开始将地球从灰色的火山岩改造成我们今天所知道的绿洲。这一事件发生在大约 24 亿年前，它发生得如此之快，至少在地质学上是如此，以至于它被称为大氧化事件。在一亿年的时间里，氧气含量猛增。不幸的是，这一事件并不是所有生命的福音。科学家们对此有一个不太宽容的名字：氧气大屠杀。
Like most processes of energy production, from burning fossil fuels to harnessing nuclear fuels, photosynthesis produced a pollutive exhaust. Instead of carbon dioxide or nuclear waste, the exhaust from photosynthesis was oxygen. Before this time, the Earth had no ozone layer. It was the cyanobacteria, with their newfound photosynthesis, that constructed Earth’s oxygen-rich atmosphere and began to terraform the planet from a gray volcanic rock to the oasis we know today. This event happened around 2.4 billion years ago, and it occurred so rapidly, at least in geological terms, that it has been called the Great Oxygenation Event. Over the course of one hundred million years, oxygen levels skyrocketed. Unfortunately, this event was not a boon for all life. Scientists have a less forgiving name for it: The Oxygen Holocaust.
氧气是一种非常活泼的元素，这使得它在细胞精心策划的化学反应中非常危险。除非采取特殊的细胞内保护措施，否则氧气化合物会干扰细胞过程，包括 DNA 的维持。这就是为什么人们认为抗氧化剂（从血液中去除高活性氧分子的化合物）可以预防癌症。光合作用生命形式成为自身成功的牺牲品，在自己的废物云中慢慢窒息。氧气的兴起带来了最致命的灾难之一地球历史上的灭绝事件。
Oxygen is an incredibly reactive element, which makes it dangerous in the carefully orchestrated chemical reactions of a cell. Unless special intracellular protective measures are taken, oxygen compounds will interfere with cellular processes, including the maintenance of DNA. This is why antioxidants—compounds that remove highly reactive oxygen molecules from your bloodstream—are believed to offer protection from cancer. The photosynthetic life-forms became victims of their own success, slowly suffocating in a cloud of their own waste. The rise of oxygen was followed by one of the deadliest extinction events in Earth’s history.
和许多危险物质（铀、汽油、煤）一样，氧气也是有用的。这种新近获得的元素提供了能量机会，生命偶然发现利用它的方法只是时间问题。一种新型细菌出现了，它不是通过光合作用而是通过细胞呼吸产生能量——氧气和糖转化为能量的过程，喷出二氧化碳作为废气。呼吸微生物开始吞噬海洋中过量的氧气，补充其耗尽的二氧化碳供应。最初对一种生命形式来说是污染物的东西，却变成了另一种生命形式的燃料。
As with many substances that are dangerous (uranium, gasoline, coal), oxygen can also be useful. This newly available element presented an energetic opportunity, and it was only a matter of time before life stumbled onto a way to exploit it. A new form of bacteria emerged that produced energy not from photosynthesis but from cellular respiration—the process by which oxygen and sugar is converted into energy, spewing out carbon dioxide as exhaust. Respiring microbes began gobbling up the ocean’s excess oxygen and replenishing its depleted supply of carbon dioxide. What began as a pollutant to one form of life became fuel for another.
地球上的生命形成了两种相互竞争但又相互补充的生命系统之间也许是有史以来最伟大的共生关系，这种共生关系一直持续到今天。一种生命是光合作用的，将水和二氧化碳转化为糖和氧气。另一种是呼吸的，将糖和氧气转化回二氧化碳。当时，这两种生命形式相似，都是单细胞细菌。今天，这种共生关系由非常不同的生命形式组成。树木、草和其他植物一些现代的生物是光合作用者，而真菌和动物则是一些现代的呼吸器。
Life on Earth fell into perhaps the greatest symbiosis ever found between two competing but complementary systems of life, one that lasts to this day. One category of life was photosynthetic, converting water and carbon dioxide into sugar and oxygen. The other was respiratory, converting sugar and oxygen back into carbon dioxide. At the time, these two forms of life were similar, both single-celled bacteria. Today this symbiosis is made up of very different forms of life. Trees, grass, and other plants are some of our modern photosynthesizers, while fungi and animals are some of our modern respirators.
图 1.2：光合作用生物与呼吸生物的共生关系
Figure 1.2: The symbiosis between photosynthetic and respiratory life
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
细胞呼吸需要糖来产生能量，而这一基本需求为最终的智力爆发提供了能量基础，这种爆发只发生在呼吸生命的后代中。虽然当时大多数（如果不是全部）微生物都表现出原始水平的智力，但只有在呼吸生命中，智力才后来得到发展和扩展。呼吸微生物与光合作用的表亲有一个关键的不同之处：它们需要狩猎。而狩猎需要全新的智慧。
Cellular respiration requires sugar to produce energy, and this basic need provided the energetic foundation for the eventual intelligence explosion that occurred uniquely within the descendants of respiratory life. While most, if not all, microbes at the time exhibited primitive levels of intelligence, it was only in respiratory life that intelligence was later elaborated and extended. Respiratory microbes differed in one crucial way from their photosynthetic cousins: they needed to hunt. And hunting required a whole new degree of smarts.
三个级别
Three Levels
二十亿年前的生态系统并不是一个饱受战争蹂躏的世界。*由能量需求促成的暂时和平支撑着生命的许多相互作用。虽然有些细菌可能吞噬了附近死去邻居的遗骸，但它们很少值得主动尝试杀死其他生命。将糖转化为能量的非氧气方法（无氧呼吸）的效率比基于氧气的方法（有氧呼吸）。因此，在引入氧气之前，狩猎不是一种可行的生存策略。最好是找到一个好地方，坐下来，沐浴在阳光下。早期生命之间最激烈的竞争可能类似于人们冲进沃尔玛参加黑色星期五大减价，为了附近稀缺的奖品而推挤他人，但不会直接攻击对方。即使是这样的推挤也可能不常见；阳光和氢气很丰富，足够用。
The ecosystem two billion years ago was not a particularly war-torn world.* A tentative peace brokered by energetic necessity undergirded life’s many interactions. Although some bacteria might have gobbled up the remains of nearby deceased neighbors, it was rarely worth the effort for them to actively try to kill other life. The non-oxygen-based approach to converting sugar into energy (anaerobic respiration) is fifteen times less efficient than the oxygen-based approach (aerobic respiration). As such, before the introduction of oxygen, hunting was not a viable survival strategy. It was better to just find a good spot, sit tight, and bask in the sunlight. The most severe competition between early life was likely akin to people rushing into a Walmart for a Black Friday sale, elbowing others out of the way for scarce nearby prizes but not directly assaulting one another. Even such elbowing was probably uncommon; sunlight and hydrogen were abundant, and there was more than enough to go around.
然而，与之前的细胞不同，呼吸生命只有通过窃取光合作用生命的能量奖赏——含糖内脏——才能生存。因此，随着有氧呼吸的出现，世界乌托邦式的和平突然结束了。正是在这里，微生物开始积极地吞食其他微生物。这为进化进程提供了动力；猎物为了避免被杀而进化出每一种防御创新，捕食者就会进化出一种进攻创新来克服同样的防御。生命陷入了军备竞赛，一个永无止境的反馈循环：进攻创新导致防御创新，而防御创新又需要进一步的进攻创新。
However, unlike the cells that came before, respiratory life could survive only by stealing the energetic prize—the sugary innards—of photosynthetic life. Thus, the world’s utopic peace ended quite abruptly with the arrival of aerobic respiration. It was here that microbes began to actively eat other microbes. This fueled the engine of evolutionary progress; for every defensive innovation prey evolved to stave off being killed, predators evolved an offensive innovation to overcome that same defense. Life became caught in an arms race, a perpetual feedback loop: offensive innovations led to defensive innovations that required further offensive innovations.
在这场混乱中，生命出现了巨大的多样化。一些物种仍然是小型单细胞微生物。其他物种进化成了第一批真核生物（发音为“you- care -ee-oats”），细胞比现在大一百倍以上，产生的能量比现在多一千倍，而且内部复杂性更高。这些真核生物是最先进的微生物杀戮机器。真核生物是第一个进化出吞噬性的生物——一种吞噬其他细胞并在其细胞壁内分解它们的狩猎策略。这些真核生物拥有更多的能量和复杂性，进一步分化为第一批植物、第一批真菌和第一批动物的前身。真核生物的真菌和动物后代保留了它们狩猎的需要（它们是呼吸器），而植物谱系则恢复了光合作用的生活方式。
Out of this maelstrom, a large diversification of life emerged. Some species remained small single-celled microbes. Other species evolved into the first eukaryotes (pronounced “you-care-ee-oats”), cells that were over a hundred times larger, produced a thousand times more energy, and had much more internal complexity. These eukaryotes were the most advanced microbial killing machines yet. Eukaryotes were the first to evolve phagotrophy—the hunting strategy of literally engulfing other cells and breaking them down inside their cellular walls. These eukaryotes, armed with more energy and complexity, further diversified into the first plants, the first fungi, and the precursors of the first animals. The fungi and animal descendants of eukaryotes retained their need to hunt (they were respirators), while plant lineage returned to a lifestyle of photosynthesis.
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
这些真核生物谱系的共同点是，植物、真菌和动物这三种生物都独立进化出了多细胞性。你所看到和认为是生命的大多数生物——人类、树木、蘑菇——主要是多细胞生物，数十亿个细胞共同作用，形成一个单一的新生生物。人类正是由这些不同类型的特化细胞组成的：皮肤细胞、肌肉细胞、肝细胞、骨细胞、免疫细胞、血细胞。植物也有特化细胞。这些细胞都发挥着不同的功能 同时仍然服务于一个共同的目的：支持整个有机体的生存。
What was common across these eukaryote lineages was that all three—plants, fungi, and animals—each independently evolved multicellularity. Most of what you see and think of as life—humans, trees, mushrooms—are primarily multicellular organisms, cacophonies of billions of individual cells all working together to create a singular emergent organism. A human is made up of exactly such diverse types of specialized cells: skin cells, muscle cells, liver cells, bone cells, immune cells, blood cells. A plant has specialized cells too. These cells all serve different functions while still serving a common purpose: supporting the survival of the overall organism.
于是，类似海藻的水下植物开始发芽，蘑菇状的真菌开始生长，原始动物开始缓慢游动。大约八亿年前，生命已经分为三个复杂的层次。第一层是单细胞生命，由微小的细菌和单细胞真核生物组成。第二层是小型多细胞生命，大到足以吞噬单细胞生物，但又小到可以使用基本的细胞螺旋桨移动。第三层是大型多细胞生命；它们太大而无法使用细胞螺旋桨移动，因此形成了无法移动的结构。
And so, seaweed-like underwater plants began sprouting, mushroom-like fungi began growing, and primitive animals began slowly swimming around. By about eight hundred million years ago, life would have fallen into three broad levels of complexity. At level one, there was single-celled life, made up of microscopic bacteria and single-celled eukaryotes. At level two, there was small multicellular life, large enough to engulf single-celled organisms but small enough to move around using basic cellular propellers. At level three, there was large multicellular life; too big to move with cellular propellers, and therefore forming immobile structures.
图 1.4：大脑出现之前的古海洋的三个复杂程度
Figure 1.4: Three complexity levels in the ancient sea before brains
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
这些早期动物可能与你想象中的动物并不相似。但它们含有与当时所有其他生命不同的元素：神经元。
These early animals probably wouldn’t resemble what you think of as animals. But they contained something that made them different from all other life at the time: neurons.
神经元
The Neuron
神经元是什么以及它有什么作用取决于你问的是谁。如果你问生物学家，神经元是构成神经系统的主要细胞。如果你问机器学习研究人员，神经元是神经网络的基本单位，是执行计算输入加权总和基本任务的小型累加器。如果你问心理物理学家，神经元是测量外部世界特征的传感器。如果你问专门研究运动控制的神经科学家，神经元是效应器，是肌肉和运动的控制器。如果你问其他人，你可能会得到各种各样的答案，从“神经元是你头脑中的小电线”到“神经元是意识的组成部分”。所有这些答案都是正确的，包含着全部真相的核心，但本身并不完整。
What a neuron is and what it does depends on whom you ask. If you ask a biologist, neurons are the primary cells that make up the nervous system. If you ask a machine learning researcher, neurons are the fundamental units of neural networks, little accumulators that perform the basic task of computing a weighted summation of their inputs. If you ask a psychophysicist, neurons are the sensors that measure features of the external world. If you ask a neuroscientist specializing in motor control, neurons are effectors, the controllers of muscles and movement. If you ask other people, you might get a wide range of answers, from “Neurons are little electrical wires in your head” to “Neurons are the stuff of consciousness.” All of these answers are right, carrying a kernel of the whole truth, but incomplete on their own.
所有动物（从蠕虫到袋熊）的神经系统都是由这些称为神经元的细长奇形细胞组成的。神经元的多样性令人难以置信，但尽管形状和大小各异，所有神经元的工作方式都相同。这是比较不同物种的神经元时最令人震惊的发现——它们大部分都是基本相同的。人类大脑中的神经元与水母中的神经元的工作方式相同。你和蚯蚓的区别不在于智力本身——神经元——而是这些单元是如何连接在一起的。
The nervous systems of all animals—from worms to wombats—are made up of these stringy odd-shaped cells called neurons. There is an incredible diversity of neurons, but despite this diversity in shapes and sizes, all neurons work the same way. This is the most shocking observation when comparing neurons across species—they are all, for the most part, fundamentally identical. The neurons in the human brain operate the same way as the neurons in a jellyfish. What separates you from an earthworm is not the unit of intelligence itself—neurons—but how these units are wired together.
图片来自 Reichert，1990 年。已获许可使用。
Figure from Reichert, 1990. Used with permission.
拥有神经元的动物有着共同的祖先，即第一个神经元进化而来的生物体。所有神经元都是从它们进化而来的。似乎在这种古老的动物祖母身上，神经元获得了现代形态；从这时起，进化重新连接了神经元，但没有对基本单位本身做出任何有意义的调整。这是一个明显的例子，说明先前的创新如何对未来的创新施加限制，通常会使早期的结构保持不变——大脑的基本构造块在六亿多年的时间里基本保持不变。
Animals with neurons share a common ancestor, an organism in whom the first neurons evolved and from whom all neurons descend. It seems that in this ancient grandmother of animals, neurons attained their modern form; from this point on, evolution rewired neurons but made no meaningful adjustments to the basic unit itself. This is a glaring example of how prior innovations impose constraints on future innovations, often leaving early structures unchanged—the fundamental building blocks of brains have been essentially the same for over six hundred million years.
为什么真菌没有神经元而动物有
Why Fungi Don’t Have Neurons but Animals Do
你和霉菌没什么不同。尽管外观不同，但真菌与动物的共同点比与植物的共同点更多。植物靠光合作用生存，而动物和真菌都靠呼吸生存。动物和真菌都呼吸氧气和吃糖；都消化食物，利用酶分解细胞并吸收其内部营养物质；而且它们都有一个共同祖先，而植物的共同祖先则要近得多，植物的分化要早得多。在多细胞生物出现的初期，真菌和动物生命应该极其相似。然而，一个谱系（动物）继续进化出神经元和大脑，而另一个谱系（真菌）却没有。为什么？
You aren’t all that different from mold. Despite their appearance, fungi have more in common with animals than they do with plants. While plants survive by photosynthesis, animals and fungi both survive by respiration. Animals and fungi both breathe oxygen and eat sugar; both digest their food, breaking cells down using enzymes and absorbing their inner nutrients; and both share a much more recent common ancestor than either do with plants, which diverged much earlier. At the dawn of multicellularity, fungi and animal life would have been extremely similar. And yet one lineage (animals) went on to evolve neurons and brains, and the other (fungi) did not. Why?
糖分只能由生命产生，因此大型多细胞呼吸生物只有两种进食方式。一种是等待生命死亡，另一种是捕捉并杀死生命。在真菌与动物分化的早期，它们各自采用了相反的进食策略。真菌选择了等待策略，而动物选择了杀死策略。*真菌人类通过外部消化进食（分泌酶在体外分解食物），而动物则通过内部消化进食（将食物困在体内，然后分泌酶）。从某些方面来看，真菌的策略比动物的策略更为成功——按生物量计算，真菌的数量大约是动物的六倍地球上的动物比人类更聪明。但正如我们不断看到的，创新往往源于更糟糕的策略、更艰难的策略。
Sugar is produced only by life, and thus there are only two ways for large multicellular respiratory organisms to feed. One is to wait for life to die, and the other is to catch and kill living life. Early in the fungi-animal divergence, they each settled into opposing feeding strategies. Fungi chose the strategy of waiting, and animals chose the strategy of killing.* Fungi eat through external digestion (secreting enzymes to break food down outside the body), while animals eat through internal digestion (trapping food inside the body and then secreting enzymes). The fungal strategy was, by some measures, more successful than the animals’—by biomass, there is about six times more fungus on Earth than animals. But as we will continually see, it is usually the worse strategy, the harder strategy, from which innovation emerges.
真菌会产生数以万亿的单细胞孢子，这些孢子漂浮在周围，处于休眠状态。如果幸运的话，一个孢子会濒临死亡，它会发育成一个巨大的真菌结构，在腐烂的组织中长出毛茸茸的细丝，分泌酶，吸收释放的营养物质。这就是为什么霉菌总是出现在过期食物中。真菌孢子在我们周围随处可见，耐心地等待着某种东西死去。真菌现在是，而且很可能一直都是地球的垃圾收集者。
Fungi produce trillions of single-celled spores that float around dormant. If by luck one happens to find itself near dying life, it will blossom into a large fungal structure, growing hairy filaments into the decaying tissue, secreting enzymes, and absorbing the released nutrients. This is why mold always shows up in old food. Fungal spores are all around us, patiently waiting for something to die. Fungi are currently, and likely have always been, Earth’s garbage collectors.
然而，早期动物采取了一种主动捕捉和吞食二级多细胞猎物的策略（见图1.4）。主动杀戮当然并不是什么新鲜事；第一批真核生物很久以前就发明了一种吞噬生命的策略。但这只对一级（单细胞）生命有效；二级多细胞团块太大，无法被单个细胞吞噬。因此，早期动物进化出了内部消化作为吞食二级生命的策略。动物独特地形成小胃，它们在里面捕获猎物，分泌酶并消化它们。
Early animals, however, settled into a strategy of actively catching and ingesting level-two multicellular prey (see figure 1.4). Active killing was, of course, not new; the first eukaryotes had long ago invented a strategy—phagotrophy—for killing life. But this worked only on level-one (single-cellular) life; level-two multicellular blobs were far too big to engulf into a single cell. And so early animals evolved internal digestion as a strategy for eating level-two life. Animals uniquely form little stomachs where they trap prey, secrete enzymes, and digest them.
事实上，形成用于消化的内腔可能是这些早期动物的决定性特征。实际上，现存的每种动物都以相同的三个初始步骤发育。从单细胞受精卵开始，形成一个空心球体（囊胚）；然后囊胚向内折叠形成一个空腔，即一个小的“胃”（原肠胚）。人类胚胎也是如此和水母胚胎一样多。虽然每种动物都是这样发育的，但没有其他生命王国这样做。这为所有动物的进化模板提供了一个明显的暗示：我们形成胃来摄取食物。所有参与这种原肠胚形成的动物也都有神经元和肌肉，似乎源自一个共同的神经元驱动的动物祖先。原肠胚形成、神经元和肌肉是将所有动物联系在一起并将动物与所有其他生命界区分开来的三个不可分割的特征。
In fact, the formation of an inner cavity for digestion may have been the defining feature of these early animals. Practically every animal alive today develops in the same three initial steps. From a single-celled fertilized egg, a hollow sphere (a blastula) forms; this then folds inward to make a cavity, a little “stomach” (a gastrula). This is true of human embryos as much as jellyfish embryos. While every animal develops this way, no other kingdom of life does this. This provides a glaring hint as to the evolutionary template on which all animals derive: we form stomachs for ingesting food. All animals that engage in such gastrulation also have neurons and muscles and seem to derive from a common neuron-enabled animal ancestor. Gastrulation, neurons, and muscles are the three inseparable features that bind all animals together and separate animals from all other kingdoms of life.
图 1.6：所有动物共同的发育阶段
Figure 1.6: Shared developmental stages for all animals
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
有些人甚至认为，动物的祖母实际上只是个小具有神经元的原肠胚状生物。但这是一个充满争议的科学领域——仅仅因为所有动物都以这种方式发育并不意味着它们真的以这种形式生活过。
Some have gone so far as to argue that the grandmother of animals was literally a little gastrula-shaped creature with neurons. But this is a scientific arena full of controversy—just because all animals develop this way doesn’t mean that they ever truly lived in this form.
另一种解释得到了化石的支持，认为第一批动物可能是类似于今天的珊瑚。肉眼看来，珊瑚非常简单，与真菌或植物并无太大区别（图 1.8）。只有仔细研究其生物学时，你才能看到动物模板的存在：胃、肌肉和神经元。珊瑚实际上是一群独立的生物，称为珊瑚虫。从某种意义上说，珊瑚虫实际上只是一个带有神经元和肌肉的胃。它们有漂浮在水中的小触手，等待小生物游向它们。当食物接触到这些触手的尖端时，它们会迅速收缩，将猎物拉入胃腔在那里，食物被消化。这些触手尖端的神经元会探测食物，并通过其他神经元网络触发信号级联，从而协调不同肌肉的放松和收缩。
Another interpretation, supported by fossils, suggests that the first animals may have been similar to today’s corals. To the naked eye, a coral is so simple that it doesn’t look all that different from a fungus or a plant (figure 1.8). Only when examining its biology closely would you see the presence of the animal template: a stomach, muscles, and neurons. A coral is actually a colony of independent organisms called coral polyps. A coral polyp is, in some sense, literally only a stomach with neurons and muscles. They have little tentacles that float in the water, waiting for small organisms to swim toward them. When food touches one of the tips of these tentacles, they rapidly contract, pulling the prey into the stomach cavity where it is digested. Neurons on the tips of these tentacles detect food and trigger a cascade of signaling through a web of other neurons that generates a coordinated relaxing and contracting of different muscles.
图 1.7：神经元赋能动物树
Figure 1.7: Tree of neuron-enabled animals
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
珊瑚的这种反射并不是多细胞生命感知和响应世界的第一个或唯一方式。植物和真菌无需神经元或肌肉就能很好地做到这一点；植物可以使叶子朝向太阳，真菌可以朝向食物生长。但是，在多细胞生命诞生之初的远古海洋中，这种反射具有革命性，这并不是因为这是多细胞生命第一次感知或移动，而是因为它是多细胞生命第一次以速度和特异性感知和移动。植物和真菌的移动需要数小时到数天；珊瑚的移动只需几秒钟。*植物和真菌的移动笨拙且不精确；珊瑚的移动相对来说非常精确。具体的——抓取猎物、张嘴、将猎物拉入胃部和闭合嘴巴都需要适时且精确的协调，放松一些肌肉，同时收缩其他肌肉。这就是为什么真菌没有神经元而动物有神经元的原因。虽然两者都是以其他生命为食的大型多细胞生物，但只有杀死二级多细胞生命的动物生存策略需要快速而具体的反射。*神经元和肌肉的最初目的可能是吞咽这项简单而又不光彩的任务。
This coral reflex was not the first or the only way in which multicellular life sensed and responded to the world. Plants and fungi do this just fine without neurons or muscles; plants can orient their leaves toward the sun, and fungi can orient their growth in the direction of food. But still, in the ancient sea at the dawn of multicellularity, this reflex would have been revolutionary, not because it was the first time multicellular life sensed or moved but because it was the first time it sensed and moved with speed and specificity. The movement of plants and fungi takes hours to days; the movement of coral takes seconds.* The movement of plants and fungi is clumsy and inexact; the movement of coral is comparatively very specific—the grasping of prey, opening of the mouth, pulling into the stomach, and closing of the mouth all require a well-timed and accurate orchestration of relaxing some muscles while contracting others. And this is why fungi don’t have neurons and animals do. Although both are large multicellular organisms that feed on other life, only the animal-survival strategy of killing level-two multicellular life requires fast and specific reflexes.* The original purpose of neurons and muscles may have been the simple and inglorious task of swallowing.
图 1.8：软珊瑚作为早期动物生命的模型
Figure 1.8: Soft coral as a model for early animal life
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
埃德加·阿德里安的三大发现和神经元的普遍特征
Edgar Adrian’s Three Discoveries and the Universal Features of Neurons
我们了解神经元如何工作的科学历程是漫长而又充满挑战的。错误的开始和错误的转折。早在公元前 400 年著名的古希腊医生希波克拉底就已知动物体内含有一种后来被称为神经（来自拉丁语nervus，意为“筋”）的纤维状物质系统，这些物质往返于大脑，是控制肌肉和感知感觉的媒介。他们通过切断活猪和其他牲畜的脊髓和钳住神经的恐怖实验知道了这一点。但古希腊人错误地认为，流经这些神经的是“动物精神”。这个错误直到几个世纪后才被纠正。两千多年后，就连伟大的艾萨克·牛顿也错误地推测，神经通过一种他称之为以太的神经液中的振动进行交流。直到 18 世纪后期，科学家才发现流经神经系统的是不是以太而是电。施加到神经上的电流将流向下游肌肉并导致其收缩。
The scientific journey by which we have come to understand how neurons work has been long and full of false starts and wrong turns. As far back as the famous ancient Greek physician Hippocrates in 400 BCE, it has been known that animals contain a system of stringy material later called nerves (from the Latin word nervus, meaning “sinew”), flowing to and from the brain, and that this system is the medium by which muscles are controlled and sensations are perceived. They knew this through horrifying experiments of severing spinal cords and clamping nerves in living pigs and other livestock. But the ancient Greeks wrongly concluded that it was “animal spirits” that flowed through these nerves. It would be many centuries before this error was corrected. More than two thousand years later, even the great Isaac Newton would speculate, incorrectly, that nerves communicated via vibrations flowing through a nerve fluid he called ether. It wasn’t until the late 1700s that scientists discovered that flowing through the nervous system was not ether but electricity. Electrical current applied to a nerve will flow toward downstream muscles and cause them to contract.
但仍然存在许多错误。当时，人们认为神经系统是由一个神经网络组成的，类似于循环系统的同质血管网。直到 19 世纪末，随着更先进的显微镜和染色技术的发展，科学家才发现神经系统是由独立的细胞（神经元）组成的，这些细胞虽然连接在一起，但却是独立的，产生自己的信号。这也表明，电信号在神经元内只沿一个方向流动，从接受输入的一部分（称为树突）流向发出电输出的另一部分（称为轴突）。该输出流向其他神经元或其他类型的细胞（例如肌肉细胞）以激活它们。
But many errors remained. At the time, it was thought that the nervous system was made up of a single web of nerves, analogous to the circulatory system’s homogenous web of blood vessels. It wasn’t until the end of the nineteenth century, with more advanced microscopes and staining techniques, that scientists discovered that the nervous system was made up of independent cells—neurons—that, although wired together, are separate and generate their own signals. This also revealed that electrical signals flow only in a single direction within a neuron, from one part that takes inputs, called dendrites, to another that sends electrical output, called the axon. This output flows to other neurons or other types of cells (such as muscle cells) to activate them.
20 世纪 20 年代初，一位名叫埃德加·阿德里安的年轻英国神经学家在第一次世界大战期间长期从事医疗服务后回到了剑桥大学。阿德里安和当时的许多研究人员一样，对用电记录神经元以破译它们如何以及它们传达什么信息很感兴趣。问题一直是，电记录设备太大太粗糙，无法记录单个神经元的活动，因此总是会产生来自多个神经元的嘈杂信号。阿德里安和他的同事们是第一个找到这个问题的技术解决方案的人，他们发明了单神经元电生理学领域。这让科学家们首次能够看到单个神经元的语言。随后的三项发现获得了阿德里安获得诺贝尔奖。
In the early 1920s, a young English neurologist by the name of Edgar Adrian returned to Cambridge University after a long stint of medical service during World War I. Adrian, like many researchers at the time, was interested in electrically recording neurons to decipher how and what they communicated. The problem had always been that electrical recording devices were too large and crude to record the activity of a single neuron and hence always produced a noisy mess of signals from multiple neurons. Adrian and his collaborators were the first to find a technical solution to this problem, inventing the field of single-neuron electrophysiology. This gave scientists, for the first time, a view into the language of individual neurons. The subsequent three discoveries earned Adrian the Nobel Prize.
第一个发现是，神经元并不以连续的起伏形式发送电信号，而是以全有或全无的反应形式发送电信号，也称为脉冲或动作电位。神经元要么打开，要么关闭；没有中间状态。换句话说，神经元的行为不像一条有恒定电流的电力线，而更像一条电报电缆，有电点击和暂停的模式。阿德里安本人注意到神经脉冲和摩尔斯电码之间的相似性。
The first discovery was that neurons don’t send electrical signals in the form of a continuous ebbing and flowing but rather in all-or-nothing responses, also called spikes or action potentials. A neuron is either on or off; there is no in between. In other words, neurons act less like an electric power line with a constant flow of electricity and more like an electric telegraph cable, with patterns of electrical clicks and pauses. Adrian himself noted the similarity between neural spikes and Morse code.
脉冲的发现给阿德里安带来了一个难题。你可以清楚地感知到感官刺激的强度——你可以区分不同音量的声音、亮度的光、气味的强度、疼痛的严重程度。一个简单的二进制信号，无论是开还是关，怎么能传达一个数值，比如刺激的等级强度？神经元的语言是脉冲，这一认识并没有让研究人员了解脉冲序列的含义。摩尔斯电码是一种代码——它是一种在一根电线上存储和传输信息的有效技巧。阿德里安是第一位使用信息这个词来指代神经元的信号，并设计了一个简单的实验来尝试解码它们。
This discovery of spikes presented a puzzle for Adrian. You can clearly perceive stimulus strengths in your senses—you can discriminate between different volumes of sound, brightness of light, potency of smells, severity of pain. How could a simple binary signal that was either on or off communicate a numerical value, such as the graded strengths of stimuli? The realization that the language of neurons was spikes didn’t tell researchers much about what a sequence of spikes meant. Morse code is, well, a code—it is an efficient trick to store and transmit information across a single electrical wire. Adrian was the first scientist to use the word information to refer to the signals of neurons, and he devised a simple experiment to try to decode them.
阿德里安从一只死去的青蛙的脖子上取下一块肌肉，并将记录设备连接到肌肉中的一个拉伸感应神经元上。这种神经元有受体，当肌肉被拉伸时会受到刺激。然后，阿德里安在肌肉上附加了各种重量。问题是：这些拉伸感应神经元的反应会如何根据施加在肌肉上的重量而变化？
Adrian took a muscle from the neck of a deceased frog and attached a recording device to a single stretch-sensing neuron in the muscle. Such neurons have receptors that are stimulated when muscles are stretched. Adrian then attached various weights to the muscle. The question was: How would the responses of these stretch-sensing neurons change based on the weight placed on the muscle?
图 1.9：Adrian 绘制了重量与这些拉伸神经元中每秒产生的脉冲数量（也就是脉冲率或发放率）之间的关系图。
Figure 1.9: Adrian charted the relationship between weight and the number of spikes per second (i.e., the spike rate, or firing rate) elicited in these stretch neurons.
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
事实证明，在所有情况下，脉冲都是相同的；唯一的区别在于发射了多少个脉冲。重量越重，脉冲的频率越高（图 1.9）。这是 Adrian 的第二个发现，现在称为速率编码。这个想法是，神经元以它们发射脉冲的速率来编码信息，而不是以形状或大小来编码脉冲本身。自 Adrian 的初始工作以来，这种速率编码已在动物界的神经元中发现，从水母到人类。触觉神经元编码压力 在其放电频率中编码对比度；光敏神经元在其放电频率中编码对比度；嗅觉敏感神经元在其放电频率中编码神经元的放电频率决定了它们的集中度。运动的神经代码也是一种放电频率：刺激肌肉的神经元的脉冲越快，收缩就越大肌肉的力量。这就是你能够温柔地抚摸你的狗并举起五十磅重的重物的原因——如果你不能调节肌肉收缩的强度，你就不会很令人愉快。
It turned out the spikes were identical in all cases; the only difference was how many spikes were fired. The heavier the weight, the higher the frequency of spikes (figure 1.9). This was Adrian’s second discovery, what is now known as rate coding. The idea is that neurons encode information in the rate that they fire spikes, not in the shape or magnitude of the spike itself. Since Adrian’s initial work, such rate coding has been found in neurons across the animal kingdom, from jellyfish to humans. Touch-sensitive neurons encode pressure in their firing rate; photosensitive neurons encode contrast in their firing rate; smell-sensitive neurons encode concentration in their firing rate. The neural code for movement is also a firing rate: the faster the spikes of the neurons that stimulate muscles, the greater the contraction force of the muscles. This is how you are able to delicately pet your dog and also pick up fifty-pound weights—if you couldn’t modulate the strength of muscle contractions, you wouldn’t be very pleasant to be around.
艾德里安的第三个发现是最令人惊讶的。试图将自然变量（例如触摸的压力或光的亮度）翻译成这种速率代码语言存在一个问题。问题是：这些自然变量的范围比神经元的放电速率所能编码的范围大得多。
Adrian’s third discovery was the most surprising of all. There is a problem with trying to translate natural variables, such as the pressure of touch or the brightness of light into this language of rate codes. The problem is this: these natural variables have a massively larger range than can be encoded in the firing rate of a neuron.
以视觉为例。你没有意识到（因为你的感觉机制把它抽象化了）你周围的光亮度变化很大。当你看东西时，进入你眼睛的光量在阳光下看白纸时，亮度比在月光下高出一百万倍。*事实上，在阳光下，一页纸上的黑色字母比白色字母亮三十倍。月光下的一页！
Take vision, for example. What you don’t realize (because your sensory machinery abstracts it away) is that the luminance of light is varying astronomically all around you. The amount of light entering your eyes when you look at a white piece of paper is one million times greater if you are in sunlight than if you are in moonlight.* In fact, the black letters of a page in sunlight are thirty times brighter than the white of a page in moonlight!
Figure 1.10: The broad spectrum of stimulus strengths
该图由 B. MacEvoy 于 2015 年制作。已获许可使用（个人通信）。
Figure made by B. MacEvoy, 2015. Used with permission (personal correspondence).
这不仅仅是光的一个特征；从嗅觉到触觉再到声音，所有感觉方式都需要辨别变化巨大的自然变量。这不一定是个问题，除非神经元存在很大的局限性——由于各种生化原因，神经元的放电速度不可能比每秒 1000 毫秒快。每秒五百个脉冲。这意味着一个神经元需要对一系列自然变量进行编码，这些变量的变化幅度可能超过一百万倍，而所有这些变量的发放率都仅为每秒 0 到 500 个脉冲。这可以合理地称为“压缩问题”：神经元必须将这一巨大的自然变量范围压缩到一个相对微小的发放率范围内。
This is not just a feature of light; all sensory modalities, from smell to touch to sound, require the discrimination of hugely varying natural variables. This wouldn’t necessarily be a problem except for a big limitation of neurons—for a variety of biochemical reasons, it is simply impossible for a neuron to fire faster than around five hundred spikes per second. This means that a neuron needs to encode a range of natural variables that can vary by a factor of over a million all within a firing rate ranging only from 0 to 500 spikes per second. This could be reasonably called the “squishing problem”: neurons have to squish this huge range of natural variables into a comparably minuscule range of firing rates.
这使得速率编码本身站不住脚。神经元根本无法在如此小的放电速率范围内直接编码如此广泛的自然变量，而不会损失大量的精度。由此产生的不精确性将使我们无法读取内部信息、检测细微的气味或注意到柔软的触摸。
This makes rate coding, on its own, untenable. Neurons simply cannot directly encode such a wide range of natural variables in such a small range of firing rates without losing a huge amount of precision. The resulting imprecision would make it impossible to read inside, detect subtle smells, or notice a soft touch.
事实证明，神经元对这个问题有一个巧妙的解决方案。神经元在自然变量和放电频率之间没有固定的关系。相反，神经元总是根据环境调整它们的放电频率；它们不断地重新映射自然界变量与放电频率语言之间的关系。神经科学家用来描述这一观察结果的术语是适应性；这是阿德里安的第三个发现。
It turns out that neurons have a clever solution to this problem. Neurons do not have a fixed relationship between natural variables and firing rates. Instead, neurons are always adapting their firing rates to their environment; they are constantly remapping the relationship between variables in the natural world and the language of firing rates. The term neuroscientists use to describe this observation is adaptation; this was Adrian’s third discovery.
在 Adrian 的青蛙肌肉实验中，一个神经元可能会在受到特定重量时发射 100 个脉冲。但在第一次接触后，神经元会迅速适应；如果你在不久后施加相同的重量，它可能只会引起 80 个脉冲。随着你继续这样做，脉冲的数量会持续下降。这适用于动物大脑中的许多神经元——刺激越强，神经元的变化就越大脉冲阈值。从某种意义上说，神经元更多的是刺激强度相对变化的测量，它表示刺激强度相对于其基线的变化程度，而不是表示刺激的绝对值。
In Adrian’s frog muscle experiments, a neuron might fire one hundred spikes in response to a certain weight. But after this first exposure, the neuron quickly adapts; if you apply the same weight shortly after, it might elicit only eighty spikes. And as you keep doing this, the number of spikes continues to decline. This applies in many neurons throughout the brains of animals—the stronger the stimuli, the greater the change in the neural threshold for spiking. In some sense, neurons are more a measurement of relative changes in stimulus strengths, signaling how much the strength of a stimulus changed relative to its baseline as opposed to signaling the absolute value of the stimulus.
图 1.11
Figure 1.11
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
美妙之处在于：适应解决了挤压问题。适应使神经元能够精确编码各种刺激强度，尽管发放率范围有限。刺激越强，让神经元做出反应所需的强度就越大下次也一样。刺激越弱，神经元就越敏感。
Here’s the beauty: Adaptation solves the squishing problem. Adaptation enables neurons to precisely encode a broad range of stimulus strengths despite a limited range of firing rates. The stronger a stimulus is, the more strength will be required to get the neuron to respond similarly next time. The weaker a stimulus is, the more sensitive neurons become.
19 世纪末和 20 世纪初，关于神经元内部工作原理的发现层出不穷。这一时期，神经科学领域涌现出一大批巨擘，他们获得了众多诺贝尔奖，其中不仅有埃德加·阿德里安，还有圣地亚哥·拉蒙·卡哈尔、查尔斯·谢林顿、亨利·戴尔，John Eccles 等人。一个重要发现是神经冲动通过突触从一个神经元传递到另一个神经元，突触是神经元之间的微小间隙。输入神经元中的尖峰触发化学物质的释放，称为神经递质，这些化学物质在纳秒内穿过突触，附着在一堆蛋白质受体上，触发离子流入目标神经元，从而改变其电荷。虽然神经元内的神经通讯是电的，但跨神经元的神经通讯是化学的。*
The late nineteenth and early twentieth centuries were filled with rich discoveries about the inner workings of neurons. A long roster of giants in neuroscience emerged in this period, leading to a platter of Nobel Prizes, not only Edgar Adrian, but also Santiago Ramón y Cajal, Charles Sherrington, Henry Dale, John Eccles, and others. One important discovery was that nerve impulses pass from one neuron to another by way of synapses, which are microscopic gaps between neurons. Spikes in the input neuron trigger the release of chemicals called neurotransmitters, which travel across the synapse over the course of nanoseconds, attach to a bunch of protein receptors, trigger ions to flow into the target neuron, and thereby change its charge. While neural communication within a neuron is electrical, across neurons, it is chemical.*
20 世纪 50 年代，约翰·埃克尔斯发现神经元主要有两种神经元有两种类型：兴奋性神经元和抑制性神经元。兴奋性神经元释放神经递质，刺激与其相连的神经元；而抑制性神经元释放神经递质，抑制与其相连的神经元。换句话说，兴奋性神经元会触发其他神经元的脉冲，而抑制性神经元会抑制其他神经元的脉冲。
In the 1950s, John Eccles discovered that neurons come in two main varieties: excitatory neurons and inhibitory neurons. Excitatory neurons release neurotransmitters that excite neurons they connect to, while inhibitory neurons release neurotransmitters that inhibit neurons they connect to. In other words, excitatory neurons trigger spikes in other neurons, while inhibitory neurons suppress spikes in other neurons.
图 1.12
Figure 1.12
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
神经元的这些特征——全有或全无的脉冲、速率编码、适应性以及具有兴奋性和抑制性神经递质的化学突触——在所有动物中都是普遍存在的，即使是没有大脑的动物，如珊瑚虫和水母。为什么所有神经元都具有这些特征？如果早期动物实际上就像今天的珊瑚和海葵一样，那么神经元的这些方面使古代动物能够以快速和特异性成功地对环境做出反应，而这对于主动捕获和杀死二级多细胞生命来说是必不可少的。全有或全无的电脉冲触发快速而有组织的反射运动，因此动物可以响应最细微的触摸或气味来捕捉猎物。速率编码使动物能够根据触摸或气味的强度来调整它们的反应。适应使动物能够调整产生脉冲时的感官阈值，使它们对最细微的触摸或气味高度敏感，同时还可以防止在更高强度的刺激下过度刺激。
These features of neurons—all-or-nothing spikes, rate coding, adaptation, and chemical synapses with excitatory and inhibitory neurotransmitters—are universal across all animals, even in animals that have no brain, such as coral polyps and jellyfish. Why do all neurons share these features? If early animals were, in fact, like today’s corals and anemones, then these aspects of neurons enabled ancient animals to successfully respond to their environment with speed and specificity, something that had become necessary to actively capture and kill level-two multicellular life. All-or-nothing electrical spikes triggered rapid and orchestrated reflexive movements so animals could catch prey in response to even the subtlest of touches or smells. Rate coding enabled animals to modify their responses based on the strengths of a touch or smell. Adaptation enabled animals to adjust the sensory threshold for when spikes are generated, allowing them to be highly sensitive to even the subtlest of touches or smells while also preventing overstimulation at higher strengths of stimuli.
那么抑制神经元呢？它们为什么会进化呢？想想珊瑚虫张嘴或闭嘴这个简单任务。为了张开嘴，一组肌肉必须收缩另一只必须放松，另一只则必须闭嘴。兴奋神经元和抑制神经元的存在使得第一个神经回路能够实现反射工作所需的逻辑形式。它们可以执行“做这个，不要那个”的规则，这也许是神经元回路中出现的第一丝智慧。“做这个，不要那个”的逻辑并不新鲜——这种逻辑在单细胞的蛋白质级联中已经存在。但这种能力在神经元的媒介中得到了重现，这使得这种逻辑在三级多细胞生命的规模上成为可能。抑制神经元使捕捉和捕捉所需的内部逻辑成为可能。吞咽反射开始起作用。
What about inhibitory neurons? Why did they evolve? Consider the simple task of a coral polyp opening or closing its mouth. For its mouth to open, one set of muscles must contract and another must relax. And the converse for closing its mouth. The existence of both excitatory and inhibitory neurons enabled the first neural circuits to implement a form of logic required for reflexes to work. They can enforce the rule of “do this, not that,” which was perhaps the first glimmer of intellect to emerge from circuits of neurons. “Do this, not that” logic was not new—such logic already existed in the protein cascades of single cells. But this ability was recapitulated in the medium of neurons, which made such logic newly possible on the scale of level-three multicellular life. Inhibitory neurons enabled the necessary inner logic for catch-and-swallow reflexes to work.
图 1.13：第一个神经回路
Figure 1.13: The first neural circuit
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
虽然最早的动物，无论是原肠胚还是息肉状的生物，显然都有神经元，但它们没有大脑。就像今天的珊瑚虫和水母一样，它们的神经系统就是科学家所说的神经网络：一个由独立神经回路组成的分布式网络，可以实现它们自己的独立反射。
While the first animals, whether gastrula-like or polyp-like creatures, clearly had neurons, they had no brain. Like today’s coral polyps and jellyfish, their nervous system was what scientists call a nerve net: a distributed web of independent neural circuits implementing their own independent reflexes.
但随着捕食者-猎物进化反馈回路的全面启动、动物主动狩猎的生态位以及神经元的构建模块的出现，进化出现第一个突破只是时间问题，这导致神经网络重新连接到大脑。我们的故事真正开始于此，但它的开始方式可能与你想象的不同。
But with the evolutionary feedback loop of predator-prey in full force, with the animal niche of active hunting, and the building blocks of neurons in place, it was only a matter of time before evolution stumbled on breakthrough #1, which led to rewiring nerve nets into brains. It is here where our story truly begins, but it doesn’t begin in the way you might expect.
突破＃1
Breakthrough #1
转向和第一批两侧对称动物
Steering and the First Bilaterians
6亿年前的大脑
Your brain 600 million years ago
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
2
2
善与恶的诞生
The Birth of Good and Bad
大自然将人类置于痛苦和快乐两个至高无上的主人的统治之下。
Nature has placed mankind under the governance of two sovereign masters, pain and pleasure.
—杰里米·边沁，《道德与立法原则导论》
—JEREMY BENTHAM, AN INTRODUCTION TO THE PRINCIPLES OF MORALS AND LEGISLATION
乍一看，动物界的多样性令人惊叹——从蚂蚁到鳄鱼，从蜜蜂到狒狒，从甲壳类动物到猫，动物似乎千差万别。但如果你进一步思考，你很容易得出这样的结论：动物界的惊人之处在于多样性太少。地球上几乎所有动物的身体结构都是一样的。它们都有一个前部，里面有嘴巴、大脑和主要的感觉器官（如眼睛和耳朵），都有一个后部，用来排泄废物。
AT FIRST GLANCE, the diversity of the animal kingdom appears remarkable—from ants to alligators, bees to baboons, and crustaceans to cats, animals seem varied in countless ways. But if you pondered this further, you could just as easily conclude that what is remarkable about the animal kingdom is how little diversity there is. Almost all animals on Earth have the same body plan. They all have a front that contains a mouth, a brain, and the main sensory organs (such as eyes and ears), and they all have a back where waste comes out.
进化生物学家将具有这种身体结构的动物称为两侧对称动物，因为它们具有双侧对称性。这与我们的远亲——珊瑚虫、海葵和水母——形成了鲜明对比。这些动物的身体结构具有径向对称性；也就是说，相似的部分围绕中心轴排列，没有前后。这两类动物之间最明显的区别是它们的进食方式。两侧对称动物进食时将食物放入嘴中，然后从屁股排出废物。径向对称的动物只有一个开口——如果你愿意的话，可以称之为嘴——用来将食物吞入胃中并吐出。不可否认，两侧对称动物是两者中更合适的。
Evolutionary biologists call animals with this body plan bilaterians because of their bilateral symmetry. This is in contrast to our most distant animal cousins—coral polyps, anemones, and jellyfish—which have body plans with radial symmetry; that is, with similar parts arranged around a central axis, without any front or back. The most obvious difference between these two categories is how the animals eat. Bilaterians eat by putting food in their mouths and then pooping out waste products from their butts. Radially symmetrical animals have only one opening—a mouth-butt if you will—which swallows food into their stomachs and spits it out. The bilaterians are undeniably the more proper of the two.
图 2.1
Figure 2.1
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
图 2.2
Figure 2.2
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
人们认为最早的动物是径向对称的，但如今大多数动物都是双侧对称的。尽管现代的两侧对称动物种类繁多——从蠕虫到人类——但它们都来自生活在约 5.5 亿年前的同一侧对称动物共同祖先。为什么在这个古代动物谱系中，身体结构会从径向对称变为双侧对称？
The first animals are believed to have been radially symmetric, and yet today, most animal species are bilaterally symmetric. Despite the diversity of modern bilaterians—from worms to humans—they all descend from a single bilaterian common ancestor who lived around 550 million years ago. Why, within this single lineage of ancient animals, did body plans change from radial symmetry to bilateral symmetry?
径向对称的身体结构很适合珊瑚等待食物的策略。但是，它们却非常不适合寻找食物的狩猎策略。如果要移动，径向对称的身体结构就需要动物具有感觉机制来检测任何方向的食物位置，然后具有向任何方向移动的机制。换句话说，它们需要能够同时检测和向所有不同的方向移动。双侧对称的身体使运动变得简单得多。它们不需要一个运动系统来向任何方向移动，而只需要一个运动系统向前移动和一个运动系统转弯。双侧对称的身体不需要选择准确的方向；它们只需要选择向右还是向左调整。
Radially symmetrical body plans work fine with the coral strategy of waiting for food. But they work horribly for the hunting strategy of navigating toward food. Radially symmetrical body plans, if they were to move, would require an animal to have sensory mechanisms to detect the location of food in any direction and then have the machinery to move in any direction. In other words, they would need to be able to simultaneously detect and move in all different directions. Bilaterally symmetrical bodies make movement much simpler. Instead of needing a motor system to move in any direction, they simply need one motor system to move forward and one to turn. Bilaterally symmetrical bodies don’t need to choose the exact direction; they simply need to choose whether to adjust to the right or the left.
即使是现代人类工程师也尚未找到更好的导航结构。汽车、飞机、船只、潜艇和几乎所有人类制造的导航机器都是双侧对称的。这简直是最有效的运动系统的设计。双侧对称性允许运动装置针对单个方向（向前）进行优化，同时通过添加转动机制来解决导航问题。
Even modern human engineers have yet to find a better structure for navigation. Cars, planes, boats, submarines, and almost every human-built navigation machine is bilaterally symmetric. It is simply the most efficient design for a movement system. Bilateral symmetry allows a movement apparatus to be optimized for a single direction (forward) while solving the problem of navigation by adding a mechanism for turning.
图 2.3：为什么双侧对称更利于导航
Figure 2.3: Why bilateral symmetry is better for navigation
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
关于两侧对称动物还有另一个观察结果，也许是更重要的一个：它们是唯一有大脑的动物。这并非巧合。第一个大脑和两侧对称动物的身体具有相同的初始进化目的：它们使动物能够通过转向来导航。转向是第一个突破。
There is another observation about bilaterians, perhaps the more important one: They are the only animals that have brains. This is not a coincidence. The first brain and the bilaterian body share the same initial evolutionary purpose: They enable animals to navigate by steering. Steering was breakthrough #1.
驾驶导航
Navigating by Steering
虽然我们不知道第一批两侧对称动物究竟长什么样，但化石表明它们是无腿的蠕虫状生物，大小与一粒米那么大。有证据表明，它们最早出现在埃迪卡拉纪，即距今 6.35 亿至 5.39 亿年前。埃迪卡拉纪的海底浅水区布满了厚厚的绿色粘稠微生物垫——大量蓝藻群落沐浴在阳光下。像珊瑚、海绵和早期植物这样的感知多细胞动物在当时很常见。
Although we don’t know exactly what the first bilaterians looked like, fossils suggest they were legless wormlike creatures about the size of a grain of rice. Evidence suggests that they first emerged sometime in the Ediacaran period, an era that stretched from 635 to 539 million years ago. The seafloor of the Ediacaran was filled with thick green gooey microbial mats in its shallow areas—vast colonies of cyanobacteria basking in the sun. Sensile multicellular animals like corals, sea sponges, and early plants would have been common.
人们认为，现代线虫自早期的两侧对称动物以来一直保持相对不变；这些生物为我们了解蠕虫祖先的内部运作提供了一个窗口。线虫几乎只是两侧对称动物的基本模板：只不过是头部、嘴巴、胃、臀部、一些肌肉和大脑。
Modern nematodes are believed to have remained relatively unchanged since early bilaterians; these creatures give us a window into the inner workings of our wormlike ancestors. Nematodes are almost literally just the basic template of a bilaterian: not much more than a head, mouth, stomach, butt, some muscles, and a brain.
图 2.4：埃迪卡拉纪世界
Figure 2.4: The Ediacaran world
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
图 2.5：线虫C. elegans
Figure 2.5: The nematode C. elegans
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
最初的大脑就像线虫的大脑一样，几乎可以肯定非常简单。研究最深入的线虫，秀丽隐杆线虫，只有 302 个神经元，与人类的大脑有 850 亿个。然而，尽管线虫的大脑很小，但它们却表现出非常复杂的行为。线虫用它们极其简单的大脑所做的事，暗示了第一批两侧对称动物用它们大脑所做的事。
The first brains were, like those of nematodes, almost definitely very simple. The most well-studied nematode, Caenorhabditis elegans, has only 302 neurons, a minuscule number compared to a human’s 85 billion. However, nematodes display remarkably sophisticated behavior despite their minuscule brains. What a nematode does with their hopelessly simple brain suggests what the first bilaterians did with theirs.
线虫与珊瑚等更古老的动物之间最明显的行为差异是，线虫花费大量时间移动。这里有一个实验：将线虫放在培养皿的一侧，将一小块食物放在另一侧。三件事会显现出来：首先，它总是 找到食物。第二，它找到食物的速度比它随机游动时快得多。第三，它不会直接游向食物，而是绕圈游动在食物上。
The most obvious behavioral difference between nematodes and more ancient animals like corals is that nematodes spend a lot of time moving. Here’s an experiment: Put a nematode on one side of a petri dish, place a tiny piece of food on the other side. Three things would reveal themselves: First, it always finds the food. Second, it finds the food much faster than it would if it were simply moving around randomly. And third, it doesn’t swim directly toward the food but rather circles in on the food.
蠕虫不使用视觉；线虫看不见。它们没有眼睛，无法呈现任何可用于导航的图像。相反，蠕虫使用嗅觉。它越接近气味源，气味的浓度就越高。蠕虫利用这一事实来寻找食物。蠕虫所要做的就是转向食物颗粒浓度增加的方向，远离食物颗粒浓度减少的方向。这种导航策略非常简单但有效，非常优雅。它可以总结为两个规则：
The worm is not using vision; nematodes can’t see. They have no eyes to render any image useful for navigation. Instead, the worm is using smell. The closer it gets to the source of a smell, the higher the concentration of that smell. Worms exploit this fact to find food. All a worm must do is turn toward the direction where the concentration of food particles is increasing, and away from the direction it is decreasing. It is quite elegant how simple yet effective this navigational strategy is. It can be summarized in two rules:
图 2.6：线虫转向食物
Figure 2.6: Nematode steering toward food
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
- 如果食物气味增多，继续前进。
- If food smells increase, keep going forward.
- 如果食物气味减少，则翻转。
- If food smells decrease, turn.
这是导航领域的重大突破。事实证明，要想在复杂的海底世界中成功导航，你实际上并不需要了解这个二维世界。你不需要了解你在哪里、食物在哪里、你可能要走哪条路、可能需要多长时间，或者任何有关这个世界的有意义的东西。你所需要的只是一个大脑，它可以引导双侧身体朝着食物气味增加的方向前进，远离食物气味减少的方向。
This was the breakthrough of steering. It turns out that to successfully navigate in the complicated world of the ocean floor, you don’t actually need an understanding of that two-dimensional world. You don’t need an understanding of where you are, where food is, what paths you might have to take, how long it might take, or really anything meaningful about the world. All you need is a brain that steers a bilateral body toward increasing food smells and away from decreasing food smells.
转向不仅可用于向物体导航，还可用于远离物体。线虫具有检测光、温度和触觉的感觉细胞。它们会避开光线，因为捕食者更容易看到它们；它们会避开有害的热量和寒冷，因为这些地方它们的身体功能会变得更难发挥；它们还会避开锋利的表面，因为这些地方它们的脆弱身体可能会受伤。
Steering can be used not only to navigate toward things but to navigate away from things. Nematodes have sensory cells that detect light, temperature, and touch. They steer away from light, where predators can more easily see them; they steer away from noxious heat and cold, where their bodily functions become harder to perform; and they steer away from surfaces that are sharp, where their fragile bodies might get wounded.
这种通过转向导航的技巧并不新鲜。像细菌这样的单细胞生物以类似的方式在环境中导航。当细菌表面的蛋白质受体检测到光等刺激时，它可以触发细胞内的化学过程，改变细胞蛋白质螺旋桨的运动，从而导致其改变方向。这就是细菌等单细胞生物游向食物来源或远离危险化学物质的方式。但这种机制转向只在单个细胞的尺度上起作用，简单的蛋白质螺旋桨就能成功地重新调整整个生命体的方向。在一个包含数百万个细胞的有机体中转向需要一种全新的设置，刺激会激活神经元回路，神经元会激活肌肉细胞，从而引起特定的转向运动。因此，第一个大脑带来的突破不是转向本身，而是在多细胞生物的尺度上转向。
This trick of navigating by steering was not new. Single-celled organisms like bacteria navigate around their environments in a similar way. When a protein receptor on the surface of a bacterium detects a stimulus like light, it can trigger a chemical process within the cell that changes the movement of the cell’s protein propellers, thereby causing it to change its direction. This is how single-celled organisms like bacteria swim toward food sources or away from dangerous chemicals. But this mechanism works only on the scale of individual cells, where simple protein propellers can successfully reorient the entire life-form. Steering in an organism that contains millions of cells required a whole new setup, one in which a stimulus activates circuits of neurons and the neurons activate muscle cells, causing specific turning movements. And so the breakthrough that came with the first brain was not steering per se, but steering on the scale of multicellular organisms.
图 2.7：线虫和扁虫等简单的两侧对称动物做出的转向决策的示例。
Figure 2.7: Examples of steering decisions made by simple bilaterians like nematodes and flatworms.
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
第一个机器人
The First Robot
20 世纪 80 年代和 90 年代，人工智能界出现了分裂。一方面是符号人工智能阵营，他们专注于将人类智能分解成各个组成部分，试图将我们最珍视的技能：推理、语言、解决问题和逻辑注入人工智能系统。另一方面是行为人工智能阵营，由麻省理工学院的机器人专家 Rodney Brooks 领导，他认为符号方法注定会失败，因为“除非我们对简单级别的智能进行大量实践，否则我们永远不会理解如何分解人类级别的智能。”
In the 1980s and 1990s a schism emerged in the artificial intelligence community. On one side were those in the symbolic AI camp, who were focused on decomposing human intelligence into its constituent parts in an attempt to imbue AI systems with our most cherished skills: reasoning, language, problem solving, and logic. In opposition were those in the behavioral AI camp, led by the roboticist Rodney Brooks at MIT, who believed the symbolic approach was doomed to fail because “we will never understand how to decompose human level intelligence until we’ve had a lot of practice with simpler level intelligences.”
布鲁克斯的论证部分基于进化论：生命花了数十亿年才能够简单地感知和响应环境；大脑又花了五亿年的时间才掌握运动技能和导航；而只有在所有这些艰苦的工作之后，语言和逻辑才出现。布鲁克斯认为，与感知和运动进化所需的时间相比，逻辑和语言的出现只是一眨眼的事。因此，他得出结论：“一旦存在和反应的本质存在，语言……和推理都非常简单。这种本质是在动态环境中移动的能力，感知周围环境的能力足以实现生命和繁殖的必要维持。智力的这一部分是进化集中精力的地方——它要困难得多。”
Brooks’s argument was partly based on evolution: it took billions of years before life could simply sense and respond to its environment; it took another five hundred million years of tinkering for brains to get good at motor skills and navigation; and only after all of this hard work did language and logic appear. To Brooks, compared to how long it took for sensing and moving to evolve, logic and language appeared in a blink of an eye. Thus he concluded that “language . . . and reason, are all pretty simple once the essence of being and reacting are available. That essence is the ability to move around in a dynamic environment, sensing the surroundings to a degree sufficient to achieve the necessary maintenance of life and reproduction. This part of intelligence is where evolution has concentrated its time—it is much harder.”
对于布鲁克斯来说，虽然人类“为我们提供了[人类水平的智能]存在的证据，但我们必须谨慎对待这些经验教训，可以从中获得好处。”为了说明这一点，他打了一个比喻：
To Brooks, while humans “provided us with an existence proof of [human-level intelligence], we must be careful about what the lessons are to be gained from it.” To illustrate this, he offered a metaphor:
假设这是 19 世纪 90 年代。人工飞行是科学、工程和风险投资界的热门话题。一群 [人工飞行] 研究人员奇迹般地乘坐时光机回到了 20 世纪 90 年代几个小时。他们全程待在一架中型商用波音 747 客机的客舱中。回到 19 世纪 90 年代后，他们感到精神振奋，因为他们知道 [人工飞行] 是可以大规模实现的。他们立即着手复制他们所看到的东西。他们在设计倾斜座椅、双层玻璃窗方面取得了巨大进展，并且知道，只要他们能搞清楚那些奇怪的“塑料”，他们就能拥有圣杯就在他们手中。
Suppose it is the 1890s. Artificial flight is the glamor subject in science, engineering, and venture capital circles. A bunch of [artificial flight] researchers are miraculously transported by a time machine to the 1990s for a few hours. They spend the whole time in the passenger cabin of a commercial passenger Boeing 747 on a medium duration flight. Returned to the 1890s they feel invigorated, knowing that [artificial flight] is possible on a grand scale. They immediately set to work duplicating what they have seen. They make great progress in designing pitched seats, double pane windows, and know that if only they can figure out those weird “plastics” they will have the grail within their grasp.
通过尝试跳过简单的飞机并直接制造 747，他们冒着完全误解飞机工作原理的风险（倾斜座椅、玻璃窗和塑料是错误的关注点）。布鲁克斯认为试图对人类大脑进行逆向工程也面临同样的问题。更好的方法是“逐步建立智能系统的能力，每一步都有完整的系统。”换句话说，就像进化一样，从简单的大脑开始，然后再增加复杂性。
By trying to skip simple planes and directly build a 747, they risked completely misunderstanding the principles of how planes work (pitched seats, paned windows, and plastics are the wrong things to focus on). Brooks believed the exercise of trying to reverse-engineer the human brain suffered from this same problem. A better approach was to “incrementally build up the capabilities of intelligence systems, having complete systems at each step.” In other words, to start as evolution did, with simple brains, and add complexity from there.
许多人不同意布鲁克斯的方法，但无论你是否同意他的观点，罗德尼·布鲁克斯是第一个成功制造出商用家用机器人的人，布鲁克斯向 Rosey 迈出了第一小步。商用机器人进化的第一步与大脑进化的第一步相似。布鲁克斯也是从驾驶开始的。
Many do not agree with Brooks’s approach, but whether you agree with him or not, it was Rodney Brooks who, by any reasonable account, was the first to build a commercially successful domestic robot; it was Brooks who made the first small step toward Rosey. And this first step in the evolution of commercial robots has parallels to the first step in the evolution of brains. Brooks, too, started with steering.
1990 年，布鲁克斯与他人共同创办了一家名为 iRobot 的机器人公司，并于 2002 年推出了真空吸尘器机器人 Roomba。Roomba 是一款可以自动导航到家中吸尘的机器人。这款机器人一经推出便大获成功；如今，新型号仍在生产，iRobot 的销量也已达到超过四千万台。
In 1990, Brooks cofounded a robotics company named iRobot, and in 2002, he introduced the Roomba, a vacuum-cleaner robot. The Roomba was a robot that autonomously navigated around your house vacuuming the floor. It was an immediate hit; new models are still being produced today, and iRobot has sold over forty million units.
第一款 Roomba 和第一批两侧对称动物拥有令人惊讶的共同特性。它们都拥有极其简单的传感器——第一款 Roomba 只能检测到少数几种情况，例如当它撞到墙壁时以及当它靠近充电基座时。它们都拥有简单的大脑——它们都不会使用它们接收到的微不足道的感官输入来构建周围环境的地图或识别物体。它们都是双侧对称的——Roomba 的轮子只能让它前进和后退。要改变方向，它必须在原地转弯，然后继续向前运动。
The first Roomba and the first bilaterians share a surprising number of properties. They both had extremely simple sensors—the first Roomba could detect only a handful of things, such as when it hit a wall and when it was close to its charging base. They both had simple brains—neither used the paltry sensory input they received to build a map of their environment or to recognize objects. They both were bilaterally symmetric—the Roomba’s wheels allowed it to go forward and backward only. To change directions, it had to turn in place and then resume its forward movement.
图 2.8：Roomba。一种以与第一批两侧对称动物类似方式导航的真空吸尘机器人。
Figure 2.8: The Roomba. A vacuum-cleaning robot that navigated in a way similar to the first bilaterians.
2006 年由拉里·D·摩尔 (Larry D. Moore) 拍摄的照片。图片发布于维基百科 https://en.wikipedia.org/wiki/Roomba。
Photograph by Larry D. Moore in 2006. Picture published on Wikipedia at https://en.wikipedia.org/wiki/Roomba.
Roomba 只需随意移动，就能清扫地板的所有角落和缝隙，当它撞到障碍物时会避开它们，当电量不足时会转向充电站。每当 Roomba 撞到墙壁时，它都会随机转弯并尝试再次向前移动。当电量不足时，Roomba 会搜索来自充电站的信号，当它检测到信号时，它会转向信号最强的方向，最终回到充电站。
The Roomba could clean all the nooks and crannies of your floor by simply moving around randomly, steering away from obstacles when it bumped into them, and steering toward its charging station when it was low on battery. Whenever the Roomba hit a wall, it would perform a random turn and try to move forward again. When it was low on battery, the Roomba searched for a signal from its charging station, and when it detected the signal, it simply turned in the direction where the signal was strongest, eventually making it back to its charging station.
Roomba 和第一批两侧对称动物的导航策略并不相同。但第一个成功的家用机器人所包含的智能与第一批大脑的智能并无太大不同，这可能并非巧合。两者都使用了技巧，使它们能够在不真正理解或模拟这个世界的情况下导航复杂的世界。
The navigational strategies of the Roomba and first bilaterians were not identical. But it may not be a coincidence that the first successful domestic robot contained an intelligence not so unlike the intelligence of the first brains. Both used tricks that enabled them to navigate a complex world without actually understanding or modeling that world.
当其他人仍呆在实验室里，研究价值数百万美元的机器人时，这些机器人有眼睛、触觉和大脑，试图计算地图和运动等复杂的东西，布鲁克斯却制造了最简单的机器人，它几乎没有任何传感器，也几乎什么都不能计算。但市场就像进化一样，最看重三样东西：便宜的东西、有用的东西，以及简单到可以首先被发现的东西。
While others remained stuck in the lab working on million-dollar robots with eyes and touch and brains that attempted to compute complicated things like maps and movements, Brooks built the simplest possible robot, one that contained hardly any sensors and that computed barely anything at all. But the market, like evolution, rewards three things above all: things that are cheap, things that work, and things that are simple enough to be discovered in the first place.
虽然转向可能不会像其他智力壮举那样令人惊叹，但它确实耗能低、有效，而且简单到足以让进化的修补者偶然发现它。所以大脑就是从这里开始的。
While steering might not inspire the same awe as other intellectual feats, it was surely energetically cheap, it worked, and it was simple enough for evolutionary tinkering to stumble upon it. And so it was here where brains began.
价数和线虫大脑的内部
Valence and the Inside of a Nematode’s Brain
线虫头部周围有感觉神经元，其中一些对光有反应，另一些对触摸有反应，还有一些对特定化学物质有反应。为了使转向发挥作用，早期的两侧对称动物需要捕捉它们检测到的每种气味、触摸或其他刺激，并做出选择：我应该接近这个东西、避开这个东西还是忽略这个东西？
Around the head of a nematode are sensory neurons, some of which respond to light, others to touch, and others to specific chemicals. For steering to work, early bilaterians needed to take each smell, touch, or other stimulus they detected and make a choice: Do I approach this thing, avoid this thing, or ignore this thing?
转向方面的突破要求两侧对称动物能够将世界分为需要接近的事物（“好事物”）和需要避免的事物（“坏事物”）。甚至 Roomba 也是如此——障碍物是坏的；电池电量不足时的充电站是好的。早期的径向对称动物不会导航，所以它们从来不需要像这样对世界上的事物进行分类。
The breakthrough of steering required bilaterians to categorize the world into things to approach (“good things”) and things to avoid (“bad things”). Even a Roomba does this—obstacles are bad; charging station when low on battery is good. Earlier radially symmetric animals did not navigate, so they never had to categorize things in the world like this.
当动物将刺激分为好刺激和坏刺激时，心理学家和神经科学家称它们在为刺激赋予效价。效价是刺激的好坏。效价与道德判断无关；它是一种更为原始的东西：动物对刺激的反应是接近刺激还是避开刺激。刺激的效价当然不是客观的；化学物质、图像或温度本身没有好坏之分。相反，刺激的效价是主观的，仅由大脑对其好坏的评价来定义。
When animals categorize stimuli into good and bad, psychologists and neuroscientists say they are imbuing stimuli with valence. Valence is the goodness or badness of a stimulus. Valence isn’t about a moral judgment; it’s something far more primitive: whether an animal will respond to a stimulus by approaching it or avoiding it. The valence of a stimulus is, of course, not objective; a chemical, image, or temperature, on its own, has no goodness or badness. Instead, the valence of a stimulus is subjective, defined only by the brain’s evaluation of its goodness or badness.
线虫如何决定它所感知事物的价态？它不会先观察某物，思考它，然后决定它的价态。相反，它头部周围的感觉神经元直接发出刺激的价态信号。一组感觉神经元实际上是正价神经元；它们直接被线虫认为好的事物（如食物气味）激活。另一组感觉神经元实际上是负价神经元；它们直接被线虫认为坏的事物（如高温、捕食者气味、强光）激活。
How does a nematode decide the valence of something it perceives? It doesn’t first observe something, ponder it, then decide its valence. Instead, the sensory neurons around its head directly signal the stimulus’s valence. One group of sensory neurons are, effectively, positive-valence neurons; they are directly activated by things nematodes deem good (such as food smells). Another group of sensory neurons are, effectively, negative-valence neurons; they are directly activated by things nematodes deem bad (such as high temperatures, predator smells, bright light).
在线虫中，感觉神经元不会发出周围世界的客观特征信号——它们编码控制投票，决定线虫想要转向或远离某物的程度。在更复杂的两侧对称动物中，比如人类，并不是所有的感觉机制都是这样的——眼睛里的神经元检测图像的特征；图像的价数是在其他地方计算的。但似乎最初的大脑始于感觉神经元，它们并不关心测量世界的客观特征，而是通过简单的价数二元镜头投射整个感知。
In nematodes, sensory neurons don’t signal objective features of the surrounding world—they encode steering votes for how much a nematode wants to steer toward or away from something. In more complex bilaterians, such as humans, not all sensory machinery is like this—the neurons in your eyes detect features of images; the valence of the image is computed elsewhere. But it seems that the first brains began with sensory neurons that didn’t care to measure objective features of the world and instead cast the entirety of perception through the simple binary lens of valence.
图 2.9显示了线虫转向的简化图。价神经元通过连接到不同的下游神经元。
Figure 2.9 shows a simplified diagram of how steering works in nematodes. Valence neurons trigger different turning decisions by connecting to different downstream neurons.
考虑一下线虫如何使用这个回路寻找食物。线虫具有正价神经元，当食物气味浓度增加时，它们会触发向前运动。正如我们在早期动物神经网络中的感觉神经元中看到的那样，这些神经元可以快速适应基线水平的气味。这使得这些价神经元能够通过信号传递气味的变化。广泛的气味浓度。无论气味浓度从二到四，还是从一百到二百，这些神经元都会产生类似数量的尖峰。这使得价神经元能够不断推动线虫朝着正确的方向前进。这是“是的，继续前进！”的信号，从远处闻到食物气味一直到食物源。
Consider how a nematode uses this circuit to find food. Nematodes have positive-valence neurons that trigger forward movement when the concentration of a food smell increases. As we saw in the sensory neurons in the nerve net of earlier animals, these neurons quickly adapt to baseline levels of smells. This enables these valence neurons to signal changes across a wide range of smell concentrations. These neurons will generate a similar number of spikes whether a smell concentration goes from two to four parts or from one hundred to two hundred parts. This enables valence neurons to keep nudging the nematode in the right direction. It is the signal for Yes, keep going! from the first whiff of a faraway food smell all the way to the food source.
图 2.9：第一个大脑的接线简化图
Figure 2.9: A simplified schematic of the wiring of the first brain
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
这种适应性的运用是进化创新促成未来创新的一个例子。早期的两侧对称动物之所以能够找到食物，是因为早期的径向对称动物已经进化出了适应性。如果没有适应性，价神经元要么过于敏感（当气味太近时会不断失灵），要么不够敏感（无法检测到远处的气味）。
This use of adaptation is an example of evolutionary innovations enabling future innovations. Steering toward food in early bilaterians was possible only because adaptation had already evolved in earlier radially symmetric animals. Without adaptation, valence neurons would be either too sensitive (and continuously misfire when smells are too close) or not sensitive enough (unable to detect faraway smells).
此时，只需改变不同价神经元兴奋的条件，就可以产生新的导航行为。例如，考虑一下线虫如何导航到最佳温度。温度导航需要比简单地转向气味更聪明一些：食物气味浓度的降低总是不好的，但环境温度的降低只有当线虫已经太冷时才是坏的。如果线虫很热，那么降低温度是好事。在炎热的夏天洗个温水澡很难受，但在寒冷的冬天洗个温水澡却很舒服。最初的大脑是如何根据不同的情况以不同的方式处理温度波动的？
At this point, new navigational behaviors could emerge simply by modifying the conditions under which different valence neurons get excited. For example, consider how nematodes navigate toward optimal temperatures. Temperature navigation requires some additional cleverness relative to the simple steering toward smells: the decreasing concentration of a food smell is always bad, but the decreasing temperature of an environment is bad only if a nematode is already too cold. If a nematode is hot, then decreasing temperature is good. A warm bath is miserable in a scorching summer but heavenly in a cold winter. How did the first brains manage to treat temperature fluctuations differently depending on the context?
线虫有一个负价神经元，当温度升高时会触发转向，但只有当温度已经高于某个阈值时才会触发；这是一个“太热！”神经元。线虫还有一个“太冷！”神经元；当温度降低时会触发转向，但只有当温度已经低于某个阈值时才会触发。这两个负价神经元一起使线虫能够在温度过高时迅速避开热量，并远离当它们感觉太冷时，它们会感到寒冷。人类大脑深处有一个古老的结构，称为下丘脑，其中有以相同方式运作的温度敏感神经元。
Nematodes have a negative-valence neuron that triggers turning when temperatures increase, but only if the temperature is already above a certain threshold; it is a Too hot! neuron. Nematodes also have a Too cold! neuron; it triggers turning when temperatures decrease, but only when temperatures are already below a certain threshold. Together, these two negative-valence neurons enable nematodes to quickly steer away from heat when they’re too hot and away from cold when they’re too cold. Deep in the human brain is an ancient structure called the hypothalamus that houses temperature-sensitive neurons that work in the same way.
权衡问题
The Problem of Trade-Offs
在存在多个刺激的情况下进行转向提出了一个问题：如果不同的感觉细胞投票决定转向相反的方向，会发生什么？如果线虫同时闻到美味和危险的气味，会发生什么？
Steering in the presence of multiple stimuli presented a problem: What happens if different sensory cells vote for steering in opposite directions? What if a nematode smells both something yummy and something dangerous at the same time?
科学家曾经在这种情形下测试过线虫。将一群线虫放在培养皿的一侧，将美味的食物放在培养皿的另一侧，然后在中间放置一个危险的铜屏障（线虫讨厌铜）。然后线虫就面临一个问题：它们是否愿意穿过屏障去吃食物？令人印象深刻的是，答案是——正如你对一只稍微聪明的动物所期望的那样——视情况而定。这取决于食物气味与铜气味的相对浓度。
Scientists have tested nematodes in exactly such situations. Put a bunch of nematodes on one side of a petri dish and place yummy food on the opposite side of the petri dish, then place a dangerous copper barrier (nematodes hate copper) in the middle. Nematodes then have a problem: Are they willing to cross the barrier to get to the food? Impressively, the answer is—as you would expect from an animal with even an iota of smarts—it depends. It depends on the relative concentration of the food smell versus the copper smell.
当铜浓度较低时，大多数线虫会穿过屏障；当铜浓度中等时，只有一部分线虫会穿过屏障；当铜浓度高时，没有线虫愿意穿过屏障。
At low levels of copper, most nematodes cross the barrier; at intermediate levels of copper, only some do; at high levels of copper, no nematodes are willing to cross the barrier.
这种在决策过程中做出权衡的能力在不同种类的简单蠕虫状两侧对称动物上进行了测试，在不同感觉模式之间进行研究。研究结果一致表明，即使是最简单的大脑（神经元数量少于一千个）也能做出这些权衡。
This ability to make trade-offs in the decision making process has been tested across different species of simple wormlike bilaterians and across different sensory modalities. The results consistently show that even the simplest brains, those with less than a thousand neurons, can make these trade-offs.
图 2.10
Figure 2.10
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
这种整合各种感官输入的要求可能是转向需要大脑的原因之一，而这种大脑无法在珊瑚虫等分布式反射网络中实现。所有这些决定转向不同方向的感官输入都必须整合到一个地方才能做出一个决定；你一次只能朝一个方向前进。第一个大脑就是这个超级整合中心——一个选择转向方向的大型神经回路。
This requirement of integrating input across sensory modalities was likely one reason why steering required a brain and could not have been implemented in a distributed web of reflexes like those in a coral polyp. All these sensory inputs voting for steering in different directions had to be integrated together in a single place to make a single decision; you can go in only one direction at a time. The first brain was this mega-integration center—one big neural circuit in which steering directions were selected.
您可以从图 2.9中直观地了解其工作原理，该图显示了线虫转向电路的简化版本。正价神经元连接到触发向前运动的神经元（可称为“前进神经元”），负价神经元连接到触发转向的神经元（可称为“转向神经元”）。前进神经元积累“继续前进！”的选票，而转弯神经元积累“转身离开！”的选票。前进神经元和转弯神经元相互抑制，使该网络能够整合权衡并做出单一选择——积累更多选票的神经元获胜，并决定动物是否穿越铜屏障。
You can get an intuition for how this works from figure 2.9, which shows a simplified version of the nematode steering circuit. Positive-valence neurons connect to a neuron that triggers forward movement (what might be called a “forward neuron”), negative-valence neurons connect to a neuron that triggers turning (what might be called a “turning neuron”). The forward neuron accumulates votes for Keep going forward!, and the turning neuron accumulates votes for Turn away! The forward neuron and turn neurons mutually inhibit each other, enabling this network to integrate trade-offs and make a single choice—whichever neuron accumulates more votes wins and determines whether the animal will cross the copper barrier.
这是过去创新如何促成未来创新的另一个例子。正如两侧对称动物不能同时前进和转弯一样，珊瑚虫也不能同时张开和闭合嘴巴。早期珊瑚状动物进化出了抑制神经元，使这些相互排斥的反射能够相互竞争，以便一次只能选择一种反射；同样的机制在早期两侧对称动物身上被重新利用，使它们能够在转向决策中做出权衡。两侧对称动物不是决定张嘴还是闭嘴，而是使用抑制神经元来决定前进还是转弯。
This is another example of how past innovations enabled future innovations. Just as a bilaterian cannot both go forward and turn at the same time, a coral polyp cannot both open and close its mouth at the same time. Inhibitory neurons evolved in earlier coral-like animals to enable these mutually exclusive reflexes to compete with each other so that only one reflex could be selected at a time; this same mechanism was repurposed in early bilaterians to enable them to make trade-offs in steering decisions. Instead of deciding whether to open or close one’s mouth, bilaterians used inhibitory neurons to decide whether to go forward or turn.
你饿了吗？
Are You Hungry?
某种物质的价态取决于动物的内部状态。线虫选择是否穿过铜屏障去寻找食物，不仅取决于食物气味和铜的相对水平，还取决于线虫有多饿。如果线虫吃饱了，它就不会越过屏障去寻找食物，但如果它饿了，它就会越过屏障去寻找食物。此外，线虫可以完全改变它们的偏好，这取决于它们有多饿。如果线虫吃饱了，它就会避开二氧化碳；如果饿了，它就会靠近二氧化碳。为什么？二氧化碳是一种由食物和捕食者释放的化学物质，所以当线虫吃饱了，为了食物而追逐二氧化碳不值得冒被捕食者攻击的风险；然而，当它饿了的时候，二氧化碳发出的信号是食物而不是捕食者的信号，所以值得冒这个险。
The valence of something depends on an animal’s internal state. A nematode’s choice of whether to cross a copper barrier to get to food depends not only on the relative level of food smell and copper but also on how hungry a nematode is. It won’t cross the barrier to get to food if it is full, but it will if it is hungry. Further, nematodes can completely flip their preferences depending on how hungry they are. If a nematode is well fed, it will steer away from carbon dioxide; if hungry, it will steer toward it. Why? Carbon dioxide is a chemical that is released by both food and predators, so when a nematode is full, pursuing carbon dioxide for food isn’t worth the risk of predators; when it is hungry, however, the chance that the carbon dioxide is signaling food, not predators, makes it worth the risk.
大脑根据内部状态快速改变刺激价态的能力无处不在。比较一下在长时间没吃饭后，吃到第一口最喜欢的晚餐时唾液分泌旺盛的快感和在吃到昏迷后吃到最后一口时胃胀恶心的感觉。短短几分钟内，你最喜欢的食物就从上帝赐予人类的礼物变成了你根本不想要的东西。
The brain’s ability to rapidly flip the valence of a stimulus depending on internal states is ubiquitous. Compare the salivary ecstasy of the first bite of your favorite dinner after a long day of skipped meals to the bloated nausea of the last bite after eating yourself into a food coma. Within mere minutes, your favorite meal can transform from God’s gift to mankind to something you want nowhere near you.
实现这一目标的机制相对简单且共享两侧对称动物中都有。动物细胞在出现健康的能量水平。动物细胞在能量不足时会释放一组不同的化学物质——饥饿信号。这两种信号都会扩散到动物的全身，为动物的饥饿程度提供持久的整体信号。线虫的感觉神经元有受体，可以检测这些信号的存在并相应地改变它们的反应。秀丽隐杆线虫中的正价食物嗅觉神经元在饥饿信号存在时对食物气味的反应更灵敏，而在饱腹信号存在时反应更弱。
The mechanisms by which this occurs are relatively simple and shared across bilaterians. Animal cells release specific chemicals—“full signals” such as insulin—in response to having a healthy amount of energy. And animal cells release a different set of chemicals—hunger signals—in response to having insufficient amounts of energy. Both signals diffuse throughout an animal’s body and provide a persistent global signal for an animal’s level of hunger. The sensory neurons of nematodes have receptors that detect the presence of these signals and change their responses accordingly. Positive valence food smell neurons in C. elegans become more responsive to food smells in the presence of hunger signals and less responsive in the presence of full signals.
Roomba 中也存在内部状态。当 Roomba 充满电时，它会忽略来自其底座的信号。在这种情况下，可以说来自底座的信号具有中性价。当 Roomba 的内部状态变为电池电量低时，来自底座的信号将转变为具有正价：Roomba 将不再忽略来自其充电站的信号，并将转向充电站以补充电池电量。
Internal states are present in a Roomba as well. A Roomba will ignore the signal from its home base when it is fully charged. In this case, the signal from the home base can be said to have neutral valence. When the Roomba’s internal state changes to one where it is low on battery, the signal from home base shifts to having positive valence: the Roomba will no longer ignore the signal from its charging station and will steer toward it to replenish its battery.
转向至少需要四个要素：用于转弯的双侧身体结构、用于检测刺激并将其分为好刺激和坏刺激的价神经元、用于将输入整合为单一转向决策的大脑，以及根据内部状态调节价的能力。但进化仍在继续。早期的两侧对称动物大脑中还出现了另一种技巧，这种技巧进一步增强了转向的有效性。这种技巧是我们现在所说的情感的早期核心。
Steering requires at least four things: a bilateral body plan for turning, valence neurons for detecting and categorizing stimuli into good and bad, a brain for integrating input into a single steering decision, and the ability to modulate valence based on internal states. But still, evolution continued tinkering. There is another trick that emerged in early bilaterian brains, a trick that further bolstered the effectiveness of steering. That trick was the early kernel of what we now call emotion.
3
3
情绪的起源
The Origin of Emotion
当你听到朋友为对方政党最近的失态行为辩护时，你会感到热血沸腾，虽然这种情绪很难用语言来定义——也许是愤怒、失望、背叛和震惊的复杂混合——但显然是一种坏情绪。当你躺在温暖阳光的海滩上时，你会感到一阵阵的平静，这种感觉也很难准确定义，但显然是一种好情绪。价态不仅存在于我们对外部刺激的评估中，也存在于我们的内部状态中。
THE BLOOD-BOILING FURY you feel when you hear a friend defend the opposite political party’s most recent gaffe, while hard to define emotionally—perhaps some complex mixture of anger, disappointment, betrayal, and shock—is clearly a bad mood. The tingly serenity you feel when you lie on a warm sunny beach, also hard to define exactly, is still clearly a good mood. Valence exists not only in our assessment of external stimuli but also in our internal states.
我们的内在状态不仅具有一定程度的效价，还具有一定程度的唤醒度。血脉贲张的愤怒不仅是一种坏心情，而且是一种被唤醒的坏心情。不同于抑郁或无聊等未被唤醒的坏心情。同样，躺在温暖的海滩上那种令人心旷神怡的宁静不仅是一种好心情，而且是一种唤醒度较低的好心情。不同于被大学录取或坐过山车（如果你喜欢这类事情的话）所产生的高度唤醒的好心情。
Our internal states are not only imbued with a level of valence, but also a degree of arousal. Blood-boiling fury is not only a bad mood but an aroused bad mood. Different from an unaroused bad mood, like depression or boredom. Similarly, the tingly serenity of lying on a warm beach is not only a good mood but a good mood with low arousal. Different from the highly arousing good mood produced by getting accepted to college or riding a roller coaster (if you like that sort of thing).
情绪很复杂。定义和分类特定情绪是一件危险的事情，充满文化偏见。德语中有一个词叫sehnsucht，大致意思是想要过上不同的生活的情绪；没有直接的英语翻译。波斯语中，ænduh这个词同时表达了遗憾和悲伤的概念；达尔格瓦语中，dard这个词同时表达了焦虑和悲伤的概念。在英语中，我们有每个情绪状态都有单独的单词。哪种语言最能区分大脑产生的情绪状态的客观类别？许多职业生涯都在寻找人类大脑中的这些客观类别；今天，大多数神经科学家认为，这种客观分类确实不存在，至少在像渴望或悲伤这样的词语层面上不存在。相反，这些情绪类别似乎主要是文化习得的。我们将在以后的突破中更多地了解其工作原理。现在，我们先从询问情绪的简单起源开始。情绪的基本模板是作为一种智力技巧而发展起来的，目的是解决第一批大脑面临的一组特定问题。因此，我们从情绪最简单的两个特征开始，这两个特征不仅在人类文化中普遍存在，而且在动物界也普遍存在，这两个情绪特征是我们从第一批大脑那里继承下来的：效价和唤醒。
Emotions are complicated. Defining and categorizing specific emotions is a perilous business, ripe with cultural bias. In German, there is a word, sehnsucht, that roughly translates to the emotion of wanting a different life; there is no direct English translation. In Persian, the word ænduh expresses the concepts of regret and grief simultaneously; in Dargwa, the word dard expresses the concepts of anxiety and grief simultaneously. In English we have separate words for each. Which language best differentiates the objective categories of emotional states produced by brains? Many careers have been spent hunting for these objective categories in human brains; today, most neuroscientists believe that such objective categorizations do not exist, at least not at the level of words like sehnsucht or grief. Instead, it seems such emotion categories are largely culturally learned. We will find out more about how this works in later breakthroughs. For now, we begin by asking about the simpler origins of emotions. The basic template of emotions evolved as an intellectual trick to solve a specific set of problems faced by the first brains. And so we begin with the simplest two features of emotions, those that are universal not only across human cultures but also across the animal kingdom, those features of emotions that we inherited from the first brains: valence and arousal.
神经科学家和心理学家使用“情感”这个词来指代这些情绪的两个属性；在任何给定点，人类都处于一种情感状态，该状态由价态和唤醒这两个维度上的位置表示。虽然哲学家、心理学家和神经科学家都无法对人类情绪类别本身做出严格的定义，但情感是情绪相对公认的统一基础。
Neuroscientists and psychologists use the word affect to refer to these two attributes of emotions; at any given point, humans are in an affective state represented by a location across these two dimensions of valence and arousal. While rigorous definitions of categories of human emotions themselves elude philosophers, psychologists, and neuroscientists alike, affect is the relatively well accepted unifying foundation of emotion.
情感的普遍性可以从我们的直觉中看出；我们很容易将一系列细微的情绪词——平静、兴奋、紧张、不安、沮丧、无聊——归类为它们所衍生的情感状态（见图3.1）。情感的普遍性也可以在我们的生物学中看到。有明确的神经生理学特征可以区分唤醒水平，例如心率、出汗、瞳孔大小、肾上腺素和血压。有明确的神经生理特征可以区分效价水平，例如压力激素水平、多巴胺水平和激活特定大脑区域。尽管世界各地的文化对特定情绪类别（如愤怒和恐惧）的分类不同，但情感状态的分类却相当普遍。所有文化都有词语来传达这些概念价态和唤醒，不同文化的新生儿在唤醒和价态方面具有普遍的面部和身体特征（例如哭泣和微笑）。
The universality of affect can be seen in our intuitions; it is easy to take suites of nuanced emotion words—calm, elated, tense, upset, depressed, bored—and assign them to the affective states from which they derive (see figure 3.1). The universality of affect can also be seen in our biology. There are clear neurophysiological signatures that differentiate levels of arousal, such as heart rate, perspiration, pupil size, adrenaline, and blood pressure. And there are clear neurophysiological signatures that differentiate levels of valence, such as stress-hormone levels, dopamine levels, and the activation of specific brain regions. And while cultures around the world differ in their classifications of specific emotion categories, such as anger and fear, classifications of affective states are quite universal. All cultures have words to communicate the concepts of valence and arousal, and newborn children across cultures have universal facial and body signatures for arousal and valence (e.g., crying and smiling).
图 3.1：人类的情感状态
Figure 3.1: The affective states of humans
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
情感的普遍性超越了人类的范围，遍布整个动物界。情感是现代情绪萌发的古老种子。但情感为何会进化呢？
The universality of affect stretches beyond the bounds of humanity; it is found across the animal kingdom. Affect is the ancient seed from which modern emotions sprouted. But why did affect evolve?
黑暗中驾驶
Steering in the Dark
即使是神经系统极小的线虫也有情感状态，尽管这些状态非常简单。线虫表现出不同程度的兴奋：当它们吃饱、有压力或生病时，它们几乎不动，对外界刺激没有反应（低兴奋）；当它们饥饿、发现食物或嗅到捕食者时，它们会不断游动（高兴奋）。线虫的情感状态也表现出不同程度的效价。正效价刺激促进进食、消化和生殖活动（原始的好心情），而负效价刺激抑制所有这些活动（原始的坏心情）。
Even nematodes with their minuscule nervous systems have affective states, albeit incredibly simple ones. Nematodes express different levels of arousal: When well fed, stressed, or ill, they hardly move at all and become unresponsive to external stimuli (low arousal); when hungry, detecting food, or sniffing predators, they will continually swim around (high arousal). The affective states of nematodes also express different levels of valence. Positive-valenced stimuli facilitate feeding, digestion, and reproductive activities (a primitive good mood), while negative-valenced stimuli inhibit all of these (a primitive bad mood).
将这些不同水平的唤醒和效价放在一起，你就会得到一个原始的情感模板。负效价刺激会触发快速游动和不频繁转身的行为，这可以被认为是被唤起的坏情绪（通常称为逃避状态）的最原始版本，而发现食物会触发慢速游动和频繁转身的行为，这可以被认为是被唤起的好情绪（通常称为利用状态）的最原始版本。逃避导致线虫快速改变位置；利用导致线虫搜索其周围环境（利用周围环境寻找食物）。虽然线虫并不像人类一样拥有复杂的情感——它们不知道年轻人的爱情冲动或送孩子上大学的苦乐参半的眼泪——但它们清楚地表现出情感的基本模板。线虫这些极其简单的情感状态为情感最初进化的原因提供了线索。
Put these different levels of arousal and valence together and you get a primitive template of affect. Negative-valenced stimuli trigger a behavioral repertoire of fast swimming and infrequent turns, which can be thought of as the most primitive version of an aroused bad mood (which is often called the state of escaping), while the detection of food triggers a repertoire of slow swimming and frequent turns, which can be thought of as the most primitive version of an aroused good mood (which is often called the state of exploiting). Escaping leads worms to rapidly change location; exploiting leads worms to search their local surroundings (to exploit their surroundings for food). Although nematodes don’t share the same complexity of emotions as humans—they do not know the rush of young love or the bittersweet tears of sending a child off to college—they clearly show the basic template of affect. These incredibly simple affective states of nematodes offer a clue as to why affect evolved in the first place.
图 3.2：线虫的情感状态
Figure 3.2: The affective states of nematodes
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
假设你把一条未进食的线虫放在一个大培养皿中，其中藏有一块食物。即使你掩盖了食物的气味让线虫前往，它也不会只是呆呆地等待食物的气味。线虫会快速游动并重新定位；换句话说，它会逃跑。它这样做是因为饥饿会触发逃跑。当线虫偶然发现隐藏的食物时，它会立即放慢速度并开始快速转身，停留在它发现食物的大致位置——它会从逃跑转为利用。最终，在吃了足够的食物后，线虫会停止移动，变得无法动弹和反应迟钝。它会进入饱食状态。
Suppose you put an unfed nematode in a big petri dish with a hidden patch of food. Even if you obscure any food smell for the nematode to steer toward, the nematode won’t just dumbly sit waiting for a whiff of food. The nematode will rapidly swim and relocate itself; in other words, it will escape. It does this because one thing that triggers escape is hunger. When the nematode happens to stumble upon the hidden food, it will immediately slow down and start rapidly turning, remaining in the same general location that it found food—it will shift from escaping to exploiting. Eventually, after eating enough food, the worm will stop moving and become immobile and unresponsive. It will shift to satiation.
科学家往往避免使用情感状态这一术语来描述简单的两侧对称动物，例如线虫，而是使用更安全的术语行为状态；这避免了暗示线虫实际上有感觉。意识体验是一个哲学泥潭，我们稍后只会简要地谈到。至少在这里，这个问题可以完全回避；情感的意识体验——无论它是什么，无论它如何运作——很可能是在情感的原始潜在机制之后进化而来的。这在人类身上也可以看到——人类大脑中产生消极或积极情感状态体验的部分在进化上是较新的，并且与大脑中产生反射性回避和趋近反应的部分不同。
Scientists tend to shy away from the term affective states in simple bilaterians such as nematodes and instead use the safer term behavioral states; this avoids the suggestion that nematodes are, in fact, feeling anything. Conscious experience is a philosophical quagmire we will only briefly touch on later. Here, at least, this issue can be sidestepped entirely; the conscious experience of affect—whatever it is and however it works—likely evolved after the raw underlying mechanisms of affect. This can be seen even in humans—the parts of the human brain that generate the experience of negative or positive affective states are evolutionarily newer and distinct from the parts of the brain that generate the reflexive avoidance and approach responses.
图 3.3
Figure 3.3
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
这些情感状态的决定性特征是，尽管它们通常由外部刺激引发，但它们在刺激消失后仍会持续很长时间。情感状态的这一特征从线虫一直延伸到人类——就像线虫在嗅到捕食者一次气味后会持续数分钟处于恐惧状态一样，人类在一次不友好的社交互动后情绪会持续数小时。这种持久性的好处，至少在一开始还不清楚。如果说有什么的话，那就是它似乎有点愚蠢——即使捕食者早已离开，线虫仍会继续试图逃跑，即使经过了所有食物消失了。第一个大脑的功能是控制方向，这一观点为线虫为何具有这些状态提供了线索，而第一批两侧对称动物可能也具有这些状态：这种坚持是控制方向的必要条件。
The defining feature of these affective states is that, although often triggered by external stimuli, they persist for long after the stimuli are gone. This feature of affective states stretches all the way from nematodes to humans—just as a nematode remains in a fear-like state for many minutes after a single sniff of a predator, a human mood can be soured for hours after a single unfriendly social interaction. The benefit of this persistence, at least at first, is unclear. If anything, it seems a bit dumb—nematodes keep trying to escape even after predators are long gone and they keep trying to exploit a local area for food even after all the food is gone. The idea that the function of the first brain was for steering provides a clue as to why nematodes have—and why the first bilaterians likely also had—these states: such persistence is required for steering to work.
感官刺激，尤其是线虫检测到的简单刺激，提供的是现实世界中存在的短暂线索，而不是一致的确定性。在野外，在科学家的培养皿之外，食物不会形成完美分布的气味梯度——水流可能会扭曲甚至完全模糊的气味，扰乱了蠕虫寻找食物或远离捕食者的能力。这些持续的情感状态是克服这一挑战的一个技巧：如果我闻到一股食物的气味，但很快就消失了，那么即使我不再闻到食物的味道，附近也可能有食物。因此，在遇到食物后坚持不懈地搜寻周围环境比只在闻到食物气味时才做出反应更有效。同样，一条蠕虫穿过一片充满捕食者的区域时，不会一直闻到捕食者的气味，而是会闻到附近有捕食者的短暂气味；如果蠕虫想逃跑，即使气味已经消失，坚持游走也是一个好主意。
Sensory stimuli, especially the simple ones detected by nematodes, offer transient clues, not consistent certainties, of what exists in the real world. In the wild, outside of a scientist’s petri dish, food does not make perfectly distributed smell gradients—water currents can distort or even completely obscure smells, disrupting a worm’s ability to steer toward food or away from predators. These persistent affective states are a trick to overcome this challenge: If I detect a passing sniff of food that quickly fades, it is likely that there is food nearby even if I no longer smell it. Therefore, it is more effective to persistently search my surroundings after encountering food, as opposed to only responding to food smells in the moment that they are detected. Similarly, a worm passing through an area full of predators won’t experience a constant smell of predators but rather catch a transient hint of one nearby; if a worm wants to escape, it is a good idea to persistently swim away even after the smell has faded.
就像飞行员试图在透过不透明或模糊的窗户驾驶飞机一样，她别无选择，只能学会在黑暗中飞行，只使用外部世界闪烁的光线提供的线索。同样，蠕虫也必须进化出一种“在黑暗中驾驶”的方法——在没有感官刺激的情况下做出驾驶决定。第一个进化解决方案是情感，即可以由外部刺激触发但在刺激消失后仍会持续很长时间的行为方式。
Like a pilot trying to fly a plane while looking through an opaque or obscured window, she would have no choice but to learn to fly in the darkness, using only the clues offered by the flickers of the outside world. Similarly, worms had to evolve a way to “steer in the dark”—to make steering decisions in the absence of sensory stimuli. The first evolutionary solution was affect, behavioral repertoires that can be triggered by external stimuli but persist long after they have faded.
这种转向功能甚至出现在 Roomba 身上。事实上，出于同样的原因，Roomba 被设计成具有不同的行为状态。通常，它们通过随机移动来探索房间。但是，如果 Roomba 遇到一片污垢，它会激活污垢检测功能，从而改变其动作；它开始在局部区域内转圈。这种新的动作由检测到污垢触发，但即使在不再检测到污垢后仍会持续一段时间。为什么 Roomba 被设计成这样？因为它有效——在一个位置检测到一片污垢可以预测附近有污垢。因此，提高获取所有污垢速度的一个简单规则是在检测到污垢后转向局部搜索一段时间。这与线虫在遇到食物并局部搜索周围环境后进化为将其行为状态从探索转变为利用的原因完全相同。
This feature of steering shows up even in a Roomba. Indeed, Roombas were designed to have different behavioral states for the same reason. Normally, they explore rooms by moving around randomly. However, if a Roomba encounters a patch of dirt, it activates Dirt Detect, which changes its repertoire; it begins turning in circles in a local area. This new repertoire is triggered by the detection of dirt but persists for a time even after dirt is no longer detected. Why was the Roomba designed to do this? Because it works—detecting a patch of dirt in one location is predictive of nearby patches of dirt. Thus, a simple rule to improve the speed of getting all the dirt is to shift toward local search for a time after detecting dirt. This is exactly the same reason nematodes evolved to shift their behavioral state from exploration to exploitation after encountering food and locally search their surroundings.
多巴胺和血清素
Dopamine and Serotonin
线虫的大脑利用一种叫做神经调节剂的化学物质来产生这些情感状态。其中两种最著名的神经调节剂是多巴胺和血清素。抗抑郁药、抗精神病药、兴奋剂和迷幻药都通过操纵这些神经调节剂发挥其作用。许多精神疾病，包括抑郁症、强迫症、焦虑症、创伤后应激障碍和精神分裂症，被认为至少部分是由神经调节剂失衡引起的。神经调节剂早在人类出现之前就进化了；它们之间的联系早在第一批两侧对称动物出现时就开始发挥作用了。
The brain of a nematode generates these affective states using chemicals called neuromodulators. Two of the most famous neuromodulators are dopamine and serotonin. Antidepressants, antipsychotics, stimulants, and psychedelics all exert their effects by manipulating these neuromodulators. Many psychiatric conditions, including depression, obsessive-compulsive disorder, anxiety, post-traumatic stress disorder, and schizophrenia are believed to be caused, at least in part, by imbalances in neuromodulators. Neuromodulators evolved long before humans appeared; they began their connection to affect as far back as the first bilaterians.
兴奋性和抑制性神经元只对它们所连接的特定神经元产生特定的、短暂的影响，而神经调节神经元对许多神经元的影响则微妙、持久且广泛。不同的神经调节神经元释放不同的神经调节剂——多巴胺神经元释放多巴胺，血清素神经元释放血清素。动物大脑中的神经元对不同类型的神经调节剂都有不同类型的受体——神经调节剂可以温和地抑制某些神经元，同时激活其他神经元；它们可以使某些神经元更容易产生脉冲，而使其他神经元不太可能产生脉冲；它们可以使某些神经元对激活更敏感，同时减弱其他神经元的反应。它们甚至可以加速或减慢适应过程。将所有这些影响结合起来，这些神经调节剂可以调节整个大脑的神经活动。正是这些不同神经调节剂之间的平衡决定了线虫的情感状态。
Unlike excitatory and inhibitory neurons, which have specific, short-lived effects on only the specific neurons they connect to, neuromodulatory neurons have subtle, long-lasting, and wide-ranging effects on many neurons. Different neuromodulatory neurons release different neuromodulators—dopamine neurons release dopamine, serotonin neurons release serotonin. And neurons throughout an animal’s brain have different types of receptors for different types of neuromodulators—neuromodulators can gently inhibit some neurons while simultaneously activating others; they can make some neurons more likely to spike while making others less likely to spike; they can make some neurons more sensitive to activation while dulling the responses of others. They can even accelerate or slow down the process of adaptation. Put all these effects together, and these neuromodulators can tune the neural activity across the entire brain. It is the balance of these different neuromodulators that determines a nematode’s affective state.
线虫的多巴胺神经元从头部伸出小附属物，并含有专门用于检测食物的受体。当这些神经元检测到食物的存在时，它们会涌入大脑产生多巴胺。这会调节电路以产生剥削状态。这种影响可以持续几分钟，然后多巴胺水平再次下降，线虫回到逃逸状态。线虫体内的血清素神经元有受体，可以检测喉咙里的食物，如果有足够的血清素释放后，会引起饱腹感。
The dopamine neurons of nematodes extend small appendages out of its head and contain receptors specifically designed to detect food. When these neurons detect the presence of food, they flood the brain with dopamine. This tunes circuits to generate the state of exploitation. This effect can last for minutes before dopamine levels drop again and the nematode returns to the state of escape. Serotonin neurons in nematodes have receptors that detect the presence of food in their throats, and if enough serotonin is released, it triggers satiety.
线虫的简单大脑为我们提供了了解多巴胺和血清素的最初或至少非常早期功能的窗口。在线虫中，当检测到蠕虫周围有食物时，多巴胺就会释放，而当检测到蠕虫内部有食物时，血清素就会释放。如果多巴胺是“附近有好东西”的化学物质，那么血清素就是好事确实发生了。多巴胺会驱使人寻找食物，而血清素则会在进食后激发人的享受。
The simple brain of the nematode offers a window into the first, or at least very early, functions of dopamine and serotonin. In the nematode, dopamine is released when food is detected around the worm, whereas serotonin is released when food is detected inside the worm. If dopamine is the something-good-is-nearby chemical, then serotonin is the something-good-is-actually-happening chemical. Dopamine drives the hunt for food; serotonin drives the enjoyment of it once it is being eaten.
图 3.4：神经调节剂在早期两侧对称动物情感状态中的作用
Figure 3.4: Role of neuromodulators in affective states of first bilaterians
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
虽然多巴胺和血清素的确切功能在不同的进化谱系中得到了详细阐述，但多巴胺和血清素之间的这种基本二分法自第一批两侧对称动物出现以来就得到了显著的保留。在像线虫、蛞蝓、鱼、老鼠和人类这样不同的物种中，多巴胺是由附近的奖励释放的，并引发情感状态的唤醒和追求（开发）；而血清素是由消费奖励释放的，并引发低唤醒状态，抑制追求奖励（满足感）。当你看到你想要的东西时会发生什么，比如你饿的时候看到的食物、性感的伴侣、比赛结束时的终点线？在所有情况下，你的大脑都会释放出一阵多巴胺。当你得到你想要的东西时会发生什么，比如当你高潮、吃美味的食物，或者只是完成待办事项清单上的一项任务时？你的大脑释放血清素。
While the exact functions of dopamine and serotonin have been elaborated throughout different evolutionary lineages, this basic dichotomy between dopamine and serotonin has been remarkably conserved since the first bilaterians. In species as divergent as nematodes, slugs, fish, rats, and humans, dopamine is released by nearby rewards and triggers the affective state of arousal and pursuit (exploitation); and serotonin is released by the consumption of rewards and triggers a state of low arousal, inhibiting the pursuit of rewards (satiation). What happens when you see something you want, like food when you’re hungry, a sexy mate, the finish line at the end of a race? In all cases, your brain releases a burst of dopamine. What happens when you get something you want, like when you’re orgasming, eating delicious food, or just finishing a task on your to-do list? Your brain releases serotonin.
如果你提高老鼠大脑中的多巴胺水平，它们就会开始冲动地利用它们能找到的任何附近的奖励：暴饮暴食并试图交配不管他们见到谁，他们都会和谁在一起。如果你提高他们的血清素水平，他们就会停止进食，变得不那么冲动，而且愿意延迟满足。血清素将行为从专注于追求目标转变为通过关闭多巴胺反应来获得满足的饱腹感，并通过减弱价神经元的反应。
If you raise dopamine levels in the brain of a rat, they begin impulsively exploiting any nearby reward they can find: gorging on food and trying to mate with whomever they see. If instead you raise their serotonin levels, they stop eating and become less impulsive and more willing to delay gratification. Serotonin shifts behavior from a focused pursuit of goals to a contented satiety by turning off dopamine responses and by dulling the responses of valence neurons.
至关重要的是，所有这些神经调节神经元（如价神经元）也对内部状态敏感。当动物饥饿时，多巴胺神经元更有可能对食物线索作出反应。
And crucially, all these neuromodulatory neurons—like valence neurons—are also sensitive to internal states. Dopamine neurons are more likely to respond to food cues when an animal is hungry.
多巴胺和奖赏之间的这种联系导致多巴胺被错误地称为“快乐化学物质”。密歇根大学的神经学家肯特·贝里奇提出了一个实验范例来探索多巴胺和快乐之间的关系。老鼠和人类一样，在品尝它们喜欢的东西（例如美味的糖丸）和它们不喜欢的东西（例如苦味液体）时会做出不同的面部表情。婴儿在品尝温牛奶时会微笑，品尝苦水时会吐口水；老鼠在品尝美味的食物时会舔嘴唇，在品尝恶心的食物时会张大嘴巴摇头。贝里奇意识到他可以利用这些不同面部反应的频率来识别老鼠的快乐程度。
This connection between dopamine and reward has caused dopamine to be—incorrectly—labeled the “pleasure chemical.” Kent Berridge, a neuroscientist at the University of Michigan, came up with a experimental paradigm to explore the relationship between dopamine and pleasure. Rats, like humans, make distinct facial expressions when tasting things they like, such as yummy sugar pellets, and things they don’t like, such as bitter liquid. A baby will smile when tasting warm milk and spit when tasting bitter water; rats will lick their lips when they taste yummy food and gape their mouths and shake their heads when they taste gross food. Berridge realized he could use the frequency of these different facial reactions as a proxy for identifying pleasure in rats.
Figure 3.5: Using facial expressions to deduce pleasure (liking) and displeasure (disliking)
图片来自 Kent Berridge（私人信件）。已获许可使用。
Images from Kent Berridge (personal correspondence). Used with permission.
令许多人惊讶的是，贝里奇发现，增加大鼠大脑中的多巴胺水平不会影响它们对食物做出愉悦面部表情的程度和频率。虽然多巴胺会导致大鼠吃下大量食物，但大鼠并没有表明它们这样做是因为它们更喜欢食物。大鼠不会表现出更多愉悦的咂嘴动作。如果有的话，它们会表现出更多厌恶尽管老鼠吃得更多，但它们仍然对食物念念不忘。这就好像老鼠无法停止进食，尽管它们不再喜欢吃东西。
To the surprise of many, Berridge found that increasing dopamine levels in the brains of rats had no impact on the degree and frequency of their pleasurable facial expressions to food. While dopamine will cause rats to consume ridiculous amounts of food, the rats do not indicate they are doing so because they like the food more. Rats do not express a higher number of pleasurable lip smacks. If anything, they express more disgust with the food, despite eating more of it. It is as if rats can’t stop eating even though they no longer enjoy it.
在另一项实验中，贝里奇破坏了几只老鼠的多巴胺神经元，几乎耗尽了它们大脑中的所有多巴胺。这些老鼠会坐在大量的食物并饿死。但多巴胺的消耗对快乐没有影响；如果贝里奇把食物放进这些饥饿的老鼠嘴里，它们会表现出所有面部表情，表明人们在饥饿时吃东西时会感到的那种欣快感；它们会比以往更频繁地咂嘴。没有多巴胺，老鼠也能很好地体验快乐——它们似乎没有动力去追求它。
In another experiment, Berridge destroyed the dopamine neurons of several rats, depleting almost all the dopamine in their brains. These rats would sit next to an abundance of food and starve to death. But this dopamine depletion had no impact on pleasure; if Berridge placed food into the mouths of these hungry rats, they exhibited all the facial expressions suggesting the kind of euphoria one would feel from eating when hungry; they smacked their lips more than ever. Rats experienced pleasure just fine without dopamine—they just didn’t seem motivated to pursue it.
这一发现也已在人类身上得到证实。在 20 世纪 60 年代的一组备受争议的实验中，精神病学家罗伯特·希思在人类大脑中植入电极，这样患者就可以按下按钮来刺激自己的多巴胺神经元。患者很快就开始反复按下这个按钮，通常要按几百次每小时的次数。有人可能会认为这是因为他们“喜欢”它，但用希思的话来说：
This finding has also been confirmed in humans. In a controversial set of experiments in the 1960s, the psychiatrist Robert Heath implanted electrodes in the brains of humans so that patients could push a button to stimulate their own dopamine neurons. Patients quickly began repeatedly pressing this button, often hundreds of times an hour. One might assume this was because they “liked” it, but in Heath’s words:
患者在解释他为何如此频繁地按下隔膜按钮时表示，这种感觉就像是他正在准备性高潮。他报告说，他无法达到性高潮的终点，但他解释说，他频繁、有时疯狂地按下按钮是为了達到终点。
The patient, in explaining why he pressed the septal button with such frequency, stated that the feeling was . . . as if he were building up to a sexual orgasm. He reported that he was unable to achieve the orgastic end point, however, explaining that his frequent, sometimes frantic, pushing of the button was an attempt to reach the end point.
多巴胺本身并不是愉悦的信号，而是对未来愉悦的期待。希思的病人并没有感受到愉悦，相反，他们常常因为无法满足按钮带来的强烈渴望而极度沮丧。
Dopamine is not a signal for pleasure itself; it is a signal for the anticipation of future pleasure. Heath’s patients weren’t experiencing pleasure; to the contrary, they often became extremely frustrated at their inability to satisfy the incredible cravings the button produced.
Berridge 证明，多巴胺与喜欢事物无关，而与想要事物有关。考虑到多巴胺的进化起源，这一发现是有道理的。对于线虫来说，当它们靠近食物时会释放多巴胺，但当它们进食时不会释放。线虫中由多巴胺引发的开发行为状态——它们会放慢速度并在周围寻找食物——在许多方面是最原始的想要状态。早在第一批两侧对称动物中，多巴胺就是一种对未来好事的预期信号，而不是好事本身的信号。
Berridge proved that dopamine is less about liking things and more about wanting things. This discovery makes sense given the evolutionary origin of dopamine. In nematodes, dopamine is released when they are near food but not when they are consuming food. The dopamine-triggered behavioral state of exploitation in nematodes—in which they slow down and search their surroundings for food—is in many ways the most primitive version of wanting. As early as the first bilaterians, dopamine was a signal for the anticipation of a future good thing, not the signal for the good thing itself.
虽然多巴胺对喜欢反应没有影响，但血清素会降低喜欢和不喜欢的反应。当给老鼠服用增加血清素水平的药物时，它们对好食物咂嘴的次数减少了，对苦味食物摇头的次数也减少了。考虑到血清素的进化起源，这也是我们所期望的：血清素是一种饱腹感、现在一切都好、令人满意的化学物质，旨在关闭价态反应。
While dopamine has no impact on liking reactions, serotonin decreases both liking and disliking reactions. When given drugs that increase serotonin levels, rats smack their lips less to good food and shake their heads less to bitter food. This is also what we would expect given the evolutionary origin of serotonin: serotonin is the satiation, things-are-okay-now, satisfaction chemical, designed to turn off valence responses.
多巴胺和血清素主要参与控制情绪状态的快乐方面，即积极情绪的不同方面。还有其他同样古老的神经调节剂，它们支撑着压力、焦虑和抑郁等消极情绪的机制。
Dopamine and serotonin are primarily involved in navigating the happy side of affective states—the different flavors of positive affect. There are additional neuromodulators, equally ancient, that undergird the mechanisms of negative affect—of stress, anxiety, and depression.
当蠕虫感到压力时
When Worms Get Stressed
人类比以往任何时候都更容易患上与压力有关的疾病。每年，世界上死于自杀的人数比死于所有暴力犯罪和战争的人数加起来还要多。每年约有 80 万人自杀，有超过每年有 1500 万人试图自杀。全球有超过 3 亿人患有抑郁症——他们失去了体验快乐的能力并积极参与生活。全球有超过 2.5 亿人患有焦虑症——对周围的世界感到无端的恐惧。疾病控制中心 (CDC) 甚至为此创造了一个术语：绝望死亡。过去二十年里，绝望死亡率增加了一倍多。
Humanity suffers from stress-related diseases more than ever. More people in the world die from suicide each year than from all violent crimes and wars put together. Around 800,000 people take their lives each year, and there are over 15 million annual suicide attempts. Over 300 million people in the world suffer from depression—stripped of their ability to experience pleasure and engage in life. Over 250 million people in the world suffer from anxiety disorders—irrationally terrified of the world around them. The Centers for Disease Control (CDC) has even devised a term for this: deaths of despair. The rate of deaths of despair has more than doubled in the last twenty years.
这些人不是被狮子吃掉，也不是饿死或冻死。这些人之所以要死，是因为他们的大脑正在杀死他们。选择自杀、明知故犯地服用致命药物或暴饮暴食导致肥胖，当然都是由我们的大脑产生的行为。任何试图理解动物行为、大脑和智力本身的尝试，如果不理解这个谜题，都是不完整的：为什么进化会创造出如此灾难性且看似荒谬的缺陷的大脑？大脑的意义，与所有进化适应一样，是为了提高生存率。那么，为什么大脑会产生如此明显的自我毁灭行为呢？
These people aren’t being eaten by lions, or starving, or freezing to death. These people are dying because their brains are killing them. Choosing to commit suicide, knowingly consuming deadly drugs, or binge eating oneself into obesity are, of course, behaviors generated by our brains. Any attempt to understand animal behavior, brains, and intelligence itself is wholly incomplete without understanding this enigma: Why would evolution have created brains with such a catastrophic and seemingly ridiculous flaw? The point of brains, as with all evolutionary adaptations, is to improve survival. Why, then, do brains generate such obviously self-destructive behaviors?
线虫会迅速尝试游向新位置，这种情感逃逸状态部分是由另一类神经调节剂引发的：去甲肾上腺素、章鱼胺和肾上腺素（也称为肾上腺素）。在两侧对称动物中，包括线虫、蚯蚓、蜗牛、鱼和老鼠等不同物种，这些化学物质由负价刺激释放，并引发众所周知的战斗或逃跑反应：心率加快、血管收缩、瞳孔扩张和抑制各种奢侈活动，例如睡眠、生殖和消化。这些神经调节剂的作用部分在于直接抵消血清素的有效性——降低动物休息并且满足。
The affective state of escape, whereby nematodes rapidly attempt to swim to a new location, is in part triggered by a different class of neuromodulators: norepinephrine, octopamine, and epinephrine (also called adrenaline). Across bilaterians, including species as divergent as nematodes, earthworms, snails, fish, and mice, these chemicals are released by negative-valanced stimuli and trigger the well-known fight-or-flight response: increasing heart rate, constricting blood vessels, dilating pupils and suppressing various luxury activities, such as sleep, reproduction, and digestion. These neuromodulators work in part by directly counteracting the effectiveness of serotonin—reducing the ability of an animal to rest and be content.
即使对于线虫来说，在世界各地移动也会消耗大量能量。肾上腺素引起的逃跑反应是动物所能做出的最昂贵的行为选择之一——逃跑反应需要大量能量消耗在肌肉上，以便快速游动。因此，进化想出了一种节省能量的技巧，从而使逃跑反应持续更长时间。肾上腺素不仅会触发逃跑的行为，还会关闭一系列耗能活动，以转移注意力能量资源被输送到肌肉。糖分从全身细胞排出，细胞生长过程停止，消化暂停，生殖过程关闭，免疫系统被驯服。这被称为急性应激反应——身体对负价刺激立即做出的反应。
Moving around in the world, even for nematodes, consumes a large amount of energy. The adrenaline-induced escape response is one of the most expensive behavioral choices an animal can make—the escape response requires a large expenditure of energy on muscles for rapid swimming. So evolution came up with a trick to save energy and thereby allow the escape response to last longer. Adrenaline not only triggers the behavioral repertoire of escape; it also turns off a swath of energy-consuming activities to divert energetic resources to muscles. Sugar is expelled from cells across the body, cell growth processes are halted, digestion is paused, reproductive processes are turned off, and the immune system is tamed. This is called the acute stress response—what bodies do immediately in response to negative-valence stimuli.
但就像政府为了资助战争而出现预算赤字一样，急性应激反应会推迟基本身体机能不能无限期地持续下去。因此，我们的两侧对称动物祖先进化出了一种对压力的反调节反应——一整套抗压力化学物质，让身体为战争的结束做好准备。这些抗压力化学物质之一就是阿片类药物。
But like a government running a budget deficit to finance a war, the acute stress response’s postponement of essential bodily functions cannot go on indefinitely. Therefore, our bilaterian ancestor evolved a counterregulatory response to stress—a whole suite of anti-stress chemicals that prepare the body for the end of the war. One of these anti-stress chemicals was opioids.
罂粟不是阿片类药物的唯一来源；大脑会自行制造并释放阿片类药物应对压力源。当压力源消失，肾上腺素水平下降时，线虫不会回到基线状态。相反，残留的抗压力化学物质会启动一系列与恢复相关的过程——免疫反应、食欲和消化功能重新开启。这些缓解和恢复化学物质（如阿片类药物）部分是通过增强血清素和多巴胺信号（两者都是受到急性应激源抑制）。阿片类药物还抑制负价神经元，这有助于动物在受伤的情况下恢复和休息。当然，这就是为什么阿片类药物对所有的两侧对称动物来说都是如此有效的止痛药。阿片类药物还会使某些奢侈功能（如生殖活动）保持关闭状态，直到缓解和恢复过程完成；这就是阿片类药物会降低性欲的原因。因此，线虫、其他无脊椎动物和人类对阿片类药物都有类似的反应也就不足为奇了——长时间进食，抑制疼痛反应，并抑制生殖行为。
Poppies are not the only source of opioids; brains make their own opioids and release them in response to stressors. When the stressor goes away and adrenaline levels drop, nematodes do not go back to their baseline state. Instead, the leftover anti-stress chemicals initiate a suite of recovery-related processes—immune responses, appetite, and digestion are turned back on. These relief-and-recover chemicals like opioids do this in part by enhancing serotonin and dopamine signals (both of which are inhibited by acute stressors). Opioids also inhibit negative-valence neurons, which helps an animal recover and rest despite any injuries. This, of course, is why opioids are such potent painkillers across all bilaterians. Opioids also keep certain luxury functions, such as reproductive activities, turned off until the relief-and-recover process is done; this is why opioids decrease sex drive. It is no surprise, then, that nematodes, other invertebrates, and humans all have similar responses to opioids—prolonged bouts of feeding, inhibited pain responses, and inhibited reproductive behavior.
图 3.6：压力和抗压力激素的时间过程
Figure 3.6: The time course of stress and anti-stress hormones
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
这种缓解和恢复状态不仅仅是恢复食欲；一只饥饿 12 小时的线虫会比压力使线虫暴饮暴食。暴饮暴食后，这些之前挨饿的线虫会“昏倒”，静止不动的时间是未挨饿线虫的十倍。线虫之所以这样做，是因为压力是一种信号，表明情况很糟糕，食物可能或很快就会变得稀缺。因此，线虫会尽可能多地储存食物，为下一次饥饿做准备。早在六亿年前的第一个大脑出现的时候，压力过后暴饮暴食的系统就已经存在了。
This relief-and-recover state doesn’t merely turn appetite back on; a nematode starved for only twelve hours will eat thirty times more food than their normally hungry peers. In other words, stress makes nematodes binge food. After binging, these previously starved nematodes “pass out,” spending ten times longer in an immobile state than unstarved worms. Nematodes do this because stress is a signal that circumstances are dire and food may be, or may soon become, scarce. Thus, nematodes stock up on as much food as they can in preparation for the next experience of starvation. As far back as the first brains six hundred million years ago, the system for binge eating after a stressful experience was already put in place.
这些抗压力激素，比如阿片类药物，与多巴胺和肯特·贝里奇 (Kent Berridge) 的大鼠面部表情实验中，血清素的作用被证实。虽然多巴胺对喜欢反应没有影响，但给大鼠服用阿片类药物确实大大增加了它们对食物的喜欢反应。鉴于我们现在对阿片类药物的进化起源的了解，这是有道理的。阿片类药物是经历压力后的缓解和恢复化学物质：压力荷尔蒙会关闭正价反应（减少喜欢），但当压力源消失时，剩余的阿片类药物会重新开启这些价反应（增加喜欢）。阿片类药物让一切变得更好；它们增加喜欢反应并减少不喜欢反应；增加愉悦感并抑制疼痛。
These anti-stress hormones like opioids differed from dopamine and serotonin in Kent Berridge’s rat facial expression experiments. While dopamine had no impact on liking reactions, giving opioids to a rat did, in fact, substantially increase their liking reactions to food. This makes sense given what we now know about the evolutionary origin of opioids. Opioids are the relief-and-recover chemical after experiencing stress: stress hormones turn positive-valence responses off (decreasing liking), but when a stressor is gone, the leftover opioids turn these valence responses back on (increasing liking). Opioids make everything better; they increase liking reactions and decrease disliking reactions; increasing pleasure and inhibiting pain.
图 3.7：古老的压力循环，起源于第一批两侧对称动物
Figure 3.7: The ancient stress cycle, originating from first bilaterians
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
忧郁与忧伤
The Blahs and Blues
所有这些都描述了身体对短期压力源的反应——急性压力反应。但现代人类所遭受的压力大多来自身体对长期压力源的反应——慢性压力反应。这也起源于第一批两侧对称动物。如果线虫暴露在 30 分钟的负面刺激下（如危险的高温、极冷或有毒化学物质），它一开始会表现出急性应激反应的特征——它会试图逃跑，而应激激素会暂停身体机能。但是，仅仅两分钟没有从这种不可避免的压力中得到缓解之后，线虫就会做出一些令人惊讶的事情：它们放弃了。蠕虫停止移动；它不再试图逃跑，只是躺在那里。事实上，这种令人惊讶的行为相当聪明：只有当刺激确实可以逃脱时，花费能量逃跑才是值得的。否则，如果蠕虫通过等待来节省能量，它更有可能存活下来。进化嵌入了一种古老的生化安全措施，以确保生物体不会浪费能量试图逃离无法逃脱的东西；这种安全措施是慢性压力和抑郁的早期种子。
All this describes what bodies do in response to short-term stressors—the acute stress response. But most of the ways that stress plagues modern humanity comes from what happens to bodies in response to prolonged stressors—the chronic stress response. This, too, has origins in the first bilaterians. If a nematode is exposed to thirty minutes of a negative stimulus (such as dangerous heat, freezing cold, or toxic chemicals), at first it will exhibit the hallmarks of the acute stress response—it will try to escape, and stress hormones will pause bodily functions. But after just two minutes of no relief from this inescapable stressor, nematodes do something surprising: they give up. The worm stops moving; it stops trying to escape and just lies there. This surprising behavior is, in fact, quite clever: spending energy escaping is worth the cost only if the stimulus is in fact escapable. Otherwise, the worm is more likely to survive if it conserves energy by waiting. Evolution embedded an ancient biochemical failsafe to ensure that an organism did not waste energy trying to escape something that was inescapable; this failsafe was the early seed of chronic stress and depression.
任何持续、无法避免或重复的负面刺激，如持续的疼痛或长期的饥饿，都会使线虫的大脑变成慢性压力状态。慢性压力与急性压力并没有什么不同；压力激素和阿片类药物持续升高，长期抑制消化、免疫反应、食欲和生殖。但慢性压力至少在一个重要方面与急性压力不同：它会抑制唤醒和动机。
Any consistent, inescapable, or repeating negative stimuli, such as constant pain or prolonged starvation, will shift a nematode brain into a state of chronic stress. Chronic stress isn’t all that different from acute stress; stress hormones and opioids remain elevated, chronically inhibiting digestion, immune response, appetite, and reproduction. But chronic stress differs from acute stress in at least one important way: it turns off arousal and motivation.
慢性应激的确切生化机制，即使是在线虫中，也是复杂的，尚未完全了解。但急性应激和慢性应激状态之间似乎有一点不同，那就是慢性应激压力开始激活血清素。乍一看，这毫无意义：血清素应该是让人感到饱足和感觉良好的化学物质。但请考虑血清素的主要作用：它会关闭价神经元反应并降低唤醒度。如果将其添加到压力荷尔蒙混合物中，你会得到一种怪异但不幸的是却很熟悉的状态——麻木。这也许是最原始的抑郁症形式。当然，线虫不会像毕加索那样经历艺术上的忧郁期，它们也不一定有意识地“体验”任何东西，但线虫仍然具有从昆虫到鱼、到老鼠到人类的所有两侧对称动物抑郁发作时所发现的基本特征：价反应麻木。这会减轻疼痛，甚至使最令人兴奋的刺激也完全失去动力。心理学家将这种抑郁症的典型症状称为快感缺乏症——缺乏快乐感。
The exact biochemical mechanisms of chronic stress, even in nematodes, are complex and not fully understood. But one thing that does seem to be different between the states of acute stress and chronic stress is that chronic stress starts activating serotonin. At first glance, this makes no sense: serotonin was supposed to be the satiation and good-feels chemical. But consider the main effect of serotonin: it turns off valence neuron responses and lowers arousal. If you add this to the soup of stress hormones, you get a bizarre yet unfortunately familiar state—numbness. This is, perhaps, the most primitive form of depression. Of course, nematodes don’t go through artistic blue periods the way that Picasso did, nor do they necessarily consciously “experience” anything, but nematodes still do share the fundamental feature found in depressive episodes across all bilaterians from insects to fish, to mice, to humans: the numbing of valence responses. This dulls pain and renders even the most exciting stimuli entirely unmotivating. Psychologists call this canonical symptom of depression anhedonia—the lack (an) of pleasure (hedonia).
线虫等动物的快感缺乏似乎是一种保持能量不可避免的压力源的存在。动物不再对压力源、好的食物气味或附近的配偶做出反应。在人类中，这个古老的系统剥夺了患者体验快乐和动力的能力。这就是抑郁症的无聊或忧郁。和所有情感状态一样，慢性压力在消极情绪消退后仍然存在刺激消失后，动物不再试图逃避负面刺激，这种习得性无助甚至在许多两侧对称动物中都可以看到，包括蟑螂、蛞蝓和果蝇。
Anhedonia in animals like nematodes seems to be a trick to preserve energy in the presence of inescapable stressors. Animals no longer respond to stressors, good food smells, or nearby mates. In humans, this ancient system robs its sufferers of the ability to experience pleasure and motivation. This is the blah or blues of depression. And like all affective states, chronic stress persists after the negative stimuli have gone away. Such learned helplessness, where animals stop trying to escape from negatively valenced stimuli, is seen even in many bilaterians, including cockroaches, slugs, and fruit flies.
我们发明了可以破解这些古老系统的药物。天然阿片类药物带来的欣快感本应保留在濒死体验后的短暂时间内。但现在人类只需一粒药丸就可以不加区分地触发这种状态。这就产生了一个问题。反复向大脑注入阿片类药物会在药效消失时产生慢性压力状态——适应是不可避免的。这会让阿片类药物使用者陷入缓解、适应、慢性压力的恶性循环，需要更多的药物才能回到基线，这会导致更多的适应，从而产生更多的慢性压力。进化的限制给现代人类蒙上了一层深远的阴影。
We have invented drugs that hack these ancient systems. The euphoria provided by natural opioids is meant to be reserved for that brief period after a near-death experience. But humans can now indiscriminately trigger this state with nothing more than a pill. This creates a problem. Repeatedly flooding the brain with opioids creates a state of chronic stress when the drug wears off—adaptation is unavoidable. This then traps opioid users in a vicious cycle of relief, adaptation, chronic stress requiring more drugs to get back to baseline, which causes more adaptation and thereby more chronic stress. Evolutionary constraints cast a long shadow on modern humanity.
这些原始的情感状态在进化过程中不断传承和发展，无论我们喜欢与否，这些残余仍然是人类行为的基本基石。随着时间的推移，神经调节剂被重新用于不同的功能，并且每种情感状态都出现了新的变体。因此，虽然人类现代的情绪状态无疑比简单的二乘二价和唤醒网格更复杂、更微妙，但它们仍然保留了它们进化的基本模板的框架。
These primitive affective states were passed down and elaborated throughout evolution, and remnants are still—whether we like it or not—essential cornerstones of human behavior. Over time, neuromodulators were repurposed for different functions, and new variants of each of these affective states emerged. And so, while the modern emotional states of humans are undeniably more complex and nuanced than a simple two-by-two grid of valence and arousal, they nonetheless retain the scaffolding of the basic template from which they evolved.
虽然这些情感状态在两侧对称动物中是共通的，但我们更远的动物表亲——海葵、珊瑚和水母——却没有表现出这样的状态。这些动物中有许多甚至根本没有血清素神经元。
Although these affective states are shared across bilaterians, our more distant animal cousins—anemones, coral, and jellyfish—do not show such states. Many of these animals don’t even have serotonin neurons at all.
这让我们面临一个令人惊讶的假设：情感，尽管具有现代色彩，在 5.5 亿年前在早期的两侧对称动物中进化而来，目的仅仅是为了转向。情感的基本模板似乎源自转向中的两个基本问题。第一个是唤醒问题：我想把精力花在移动上吗？还是不？第二个是效价问题：我想留在这个位置还是离开这个位置？特定神经调节剂的释放强制对每个问题做出具体回答。然后，这些留下和离开的全局信号可用于调节一系列反射，例如产卵、交配和消耗能量消化食物是否安全。
This leaves us at the doorstep of a surprising hypothesis: Affect, despite all its modern color, evolved 550 million years ago in early bilaterians for nothing more than the mundane purpose of steering. The basic template of affect seems to have emerged from two fundamental questions in steering. The first was the arousal question: Do I want to expend energy moving or not? The second was the valence question: Do I want to stay in this location or leave this location? The release of specific neuromodulators enforced specific answers to each of these questions. And these global signals for stay and leave could then be used to modulate suites of reflexes, such as whether it was safe to lay eggs, mate, and expend energy digesting food.
然而，这些情感状态及其神经调节剂将在第一批大脑的进化过程中发挥更为基础的作用。
However, these affective states and their neuromodulators would go on to play an even more foundational role in the evolution of the first brains.
4
4
联想、预测和学习的曙光
Associating, Predicting, and the Dawn of Learning
Memory is everything. Without it we are nothing.
—埃里克·坎德尔
—ERIC KANDEL
1904 年 12 月 12 日，一位名叫伊万·巴甫洛夫的俄国科学家站在瑞典卡罗琳医学院的一群研究人员面前。两天前，巴甫洛夫成为第一位获得诺贝尔奖的俄国人。八年前，因发明炸药而致富的瑞典工程师兼商人阿尔弗雷德·诺贝尔去世，并将自己的财产捐赠给诺贝尔基金会。诺贝尔曾规定，获奖者必须就获奖主题进行演讲，因此，这一天，在斯德哥尔摩，巴甫洛夫进行了演讲。
ON DECEMBER 12, 1904, a Russian scientist by the name of Ivan Pavlov stood in front of an assembly of researchers at the Karolinska Institute in Sweden. Pavlov had, two days earlier, become the first Russian to win the Nobel Prize. Eight years prior, Alfred Nobel—the Swedish engineer and businessman who got rich from his invention of dynamite—had passed away and bequeathed his fortune to the creation of the Nobel Foundation. Nobel had stipulated that winners were to give a lecture on the subject for which the prize had been awarded, and so, on this day in Stockholm, Pavlov gave his lecture.
虽然他目前因对心理学的贡献而闻名，但这并不是他获得诺贝尔奖的原因。巴甫洛夫不是心理学家，而是生理学家——在此之前，他的整个研究生涯都在研究消化系统的潜在生物机制——即“生理学”。
Although he is currently known for his contributions to psychology, that was not the work that earned him the Nobel. Pavlov was not a psychologist but a physiologist—he had spent his entire research career up to this point studying the underlying biological mechanisms—the “physiology”—of the digestive system.
在巴甫洛夫之前，研究消化系统的唯一方法是手术切除动物的器官（食道、胃或胰腺），并在器官坏死前快速进行实验。巴甫洛夫开创了各种相对非侵入性的技术，使他能够测量特征完整健康狗的消化系统。其中最著名的是插入一个小的唾液瘘管，将唾液从一个唾液腺转移到挂在狗嘴外的小管中；这使巴甫洛夫能够确定各种刺激产生的唾液量和含量。他对食道、胃和胰腺也做了类似的实验。
Before Pavlov, the only way to study the digestive system was to surgically remove animals’ organs—esophagus, stomach, or pancreas—and run experiments quickly before the organs died. Pavlov pioneered a variety of relatively noninvasive techniques that enabled him to measure features of the digestive system in intact and healthy dogs. The most famous of these was the insertion of a small salivary fistula that diverted saliva from one salivary gland to a small tube that hung out of the dog’s mouth; this enabled Pavlov to determine the quantity and content of saliva produced by various stimuli. He did similar tricks with the esophagus, stomach, and pancreas.
通过这些新技术，巴甫洛夫和他的同事们取得了多项发现。他们了解到了人体对各种食物的反应是释放哪些类型的消化化学物质，他还发现消化器官受神经系统的控制。这些贡献使他获得了诺贝尔奖。
Through these new techniques, Pavlov and his colleagues made several discoveries. They learned what types of digestive chemicals were released in response to various foods, and he discovered that the digestive organs were under the control of the nervous system. These contributions won him the prize.
然而，演讲进行到三分之二时，巴甫洛夫就把注意力从获奖成果上转移开了。作为一名容易激动的科学家，他忍不住要介绍当时还只是推测的研究，但他相信这最终会成为他最重要的工作——他对他所谓的条件反射的探索。
However, two-thirds through his speech, Pavlov turned his focus away from his prizewinning work. An excitable scientist, he couldn’t resist pitching research that was, at the time, speculative but that he believed would eventually become his most important work—his exploration of what he called conditional reflexes.
一直以来，他都无法对消化反应进行细致的测量，因为动物在品尝食物之前，消化器官经常会受到刺激。他的狗在意识到食物实验即将开始时，就会流口水，肚子咕咕叫。这是一个问题。如果你想测量味蕾检测到肥肉或含糖水果时唾液腺的反应，你肯定不希望仅仅通过观察这些物质就测量出受试者释放的物质。
There had always been a pesky confound that got in the way of his meticulous measurements of digestive responses—digestive organs often became stimulated before animals tasted food. His dogs salivated and their stomachs gurgled the moment they realized an experiment with food was about to begin. This was a problem. If you want to measure how salivary glands respond when taste buds detect fatty meat or sugary fruit, you don’t want the confounding measurement of whatever was released by the subjects merely looking at these substances.
这种所谓的精神刺激对巴甫洛夫的研究来说是一个特别的烦恼，他认为这是一种“错误来源”。巴甫洛夫开发了各种技术来消除这种混淆；例如，实验者在单独的隔离房间中工作，以“小心地避免一切可能引发狗对食物的想法。”
This psychic stimulation, as it was called, was a particular annoyance to Pavlov’s research, what he deemed a “source of error.” Pavlov developed various techniques to eliminate this confound; for instance, experimenters worked in separate isolated rooms to “carefully avoid everything that could elicit in the dog thoughts about food.”
直到很久以后，在将心理学家带入实验室后，巴甫洛夫才开始将心理刺激视为值得分析的变量，而不是需要消除的混淆因素。讽刺的是，第一个理解心理刺激的人是一位以消除心理刺激为目标的消化生理学家。
Only much later, after bringing psychologists into his lab, did Pavlov begin to view psychic stimulation not as a confound to be eliminated but as a variable worthy of analysis. Ironically, it was a digestive physiologist with the goal of eliminating psychic stimulation who became the first to understand it.
巴甫洛夫实验室发现，精神刺激并不像似乎如此。狗会对任何刺激物（节拍器、灯光、蜂鸣器）产生唾液分泌反应，这些刺激物之前都与食物有关。如果实验者打开蜂鸣器，然后喂食，狗只会对蜂鸣器产生唾液分泌反应。狗已经形成了一种条件反射——对蜂鸣器产生唾液分泌反应的条件是蜂鸣器与食物之间的先前关联。巴甫洛夫将这些条件反射与他所谓的非条件反射进行了对比——非条件反射是天生的，不需要任何关联。饥饿的狗对放在嘴里的糖产生唾液分泌反应，无论之前是否有关联。
Pavlov’s lab discovered that psychic stimulation was not as random as it seemed. Dogs would salivate in response to any stimuli—metronomes, lights, buzzers—that had been previously associated with food. If an experimenter turned on a buzzer and then gave food, the dog began to salivate in response to the buzzer alone. The dog had developed a conditional reflex—the reflex to salivate in response to the buzzer was conditional on the prior association between the buzzer and food. Pavlov contrasted these conditional reflexes with what he called unconditional reflexes—those that were innate and required no association. A hungry dog’s reflex to salivate in response to sugar placed in its mouth occurred regardless of any prior associations.
巴甫洛夫实验后不久，其他科学家开始在其他反射上尝试这些技术。结果表明，大多数（如果不是全部）反射都会建立这样的关联。将任意声音与电击手配对，很快你的手就会缩回，只听到声音。将任意声音与轻轻吹向某人眼睛的气流配对，最终他或她会不由自主地眨眼，只听到声音。将任意声音与反射锤击某人膝盖配对，最终他们的腿会仅对声音作出反应即可。
Shortly after Pavlov’s experiments, other scientists began trying these techniques on other reflexes. It turned out that most, if not all, reflexes build such associations. Pair an arbitrary sound with an electric shock to your hand, and soon your hand will retract to just the sound. Pair an arbitrary sound with a gentle puff of air to someone’s eyes, and eventually he or she will involuntarily blink in response to just the sound. Pair an arbitrary sound with a reflex hammer to a person’s knee, and eventually their leg will kick in response to just the sound.
巴甫洛夫条件反射的典型特征是，它们是无意识的联想；人们会情不自禁地眨眼、踢腿或缩回手。就像从战场归来的士兵听到巨大的噪音会情不自禁地跳起来，或者有演讲恐惧症的人在上台前会情不自禁地紧张一样，巴甫洛夫的狗也会情不自禁地在蜂鸣器响起时流口水。巴甫洛夫条件反射的非自愿性质，即联想学习会自动发生而无需有意识的参与，是学习和记忆可能比以前认为的更古老的第一个线索。学习可能不需要进化后期出现的所有大脑结构。事实上，即使是一只被切除了整个大脑的老鼠也会表现出条件反射。如果你把拍打它的腿（导致腿缩回）和拍打它的尾巴（导致尾巴缩回）配对，那么尽管没有大脑，它也会学会在拍打腿时缩回尾巴。老鼠可以只用大脑就能学会这种联想脊髓中的简单回路。
The defining feature of Pavlov’s conditional reflexes is that they are involuntary associations; people can’t help but blink, kick, or retract their hands. Just as a soldier who returns from war cannot help but jump when they hear a loud noise or a person with a phobia of public speaking cannot help but tense up before going onstage, so too could Pavlov’s dogs not help but salivate in response to the buzzer. The involuntary nature of Pavlov’s conditional reflexes, the fact that associative learning occurs automatically without conscious involvement, was the first clue that learning and memory might be more ancient than previously thought. Learning might not require all the brain structures that emerged later in evolution. Indeed, even a rat with its entire brain removed exhibits conditional reflexes. If you pair a tap to its leg (which causes leg retraction) with a tap to its tail (which causes tail retraction), then it will learn to retract its tail in response to a leg tap, despite lacking a brain. Rats can learn this association with nothing but the simple circuits in their spinal cords.
如果联想学习是简单神经元回路的特性，即使是存在于大脑之外的神经元回路，那么它可能是一种非常古老的进化的诀窍。事实上，巴甫洛夫无意中发现了学习本身的进化起源。
If associative learning is a property of simple circuits of neurons, even those present outside of the brain, then it might be a very ancient evolutionary trick. Indeed, Pavlov had unintentionally stumbled on the evolutionary origin of learning itself.
调整事物的善与恶
Tweaking the Goodness and Badness of Things
假设你拿了一百条线虫，一半放在一个装有白水的盘子里，另一半放在一个装有盐水的盘子里。几个小时后，这些线虫会饿得难受，因为两个盘子里都没有食物。这时，把这两组线虫放进另一个盘子里，盘子的一边放了一点盐。会发生什么？
Suppose you took a hundred nematodes, put half of them in a dish with plain water and the other half in a dish with salty water. After several hours, these nematodes will become uncomfortably hungry, as neither dish contains any food. At this point, put both groups of nematodes into another dish that contains a little morsel of salt on one side. What happens?
在白水中经历过饥饿的线虫会像普通线虫一样表现：它们会转向盐（线虫通常认为盐具有正价）。然而，在盐水中经历过饥饿的线虫会做完全相反的事情：它们会转向远离盐。
The nematodes that experienced hunger with plain water will behave as normal nematodes do: they will steer toward the salt (nematodes typically consider salt to have positive valence). However, the nematodes that experienced hunger in salt water will do the exact opposite: they will steer away from the salt.
图 4.1：当盐与饥饿相关时，线虫学会避开盐
Figure 4.1: Nematodes learn to steer away from salt when salt is associated with hunger
人物形象由 Max Bennett 设计（部分图标来自 Rebecca Gelernter）
Figure designed by Max Bennett (with some icons from Rebecca Gelernter)
盐通过与饥饿的负价状态相关，从正价刺激物转变为负价刺激物。*
Salt was transformed from a positive valence stimulus to a negative valence stimulus through its association with the negative valence state of hunger.*
事实证明，巴甫洛夫的联想学习是所有两侧对称动物都具备的智力能力，即使是简单的两侧对称动物也是如此。如果你同时让线虫接触美味的食物气味和使它们生病的有毒化学物质，线虫随后就会避开 食物的气味。如果你在特定温度下喂养线虫，它们会改变它们的偏好向那个温度靠拢。轻轻拍打蛞蝓的一侧，再加上一点电击，就会引发退缩反射，蛞蝓就会学会只对拍打动作退缩，这种联想将持续数天。
It turns out that Pavlov’s associative learning is an intellectual ability of all bilaterians, even simple ones. If you expose nematodes simultaneously to both a yummy food smell and a noxious chemical that makes them sick, nematodes will subsequently steer away from that food smell. If you feed nematodes at a specific temperature, they will shift their preferences toward that temperature. Pair a gentle tap to the side of a slug with a small electric shock, which triggers a withdrawal reflex, and the slug will learn to withdraw to just the tap, an association that will last for days.
然而，虽然在两侧对称动物中都存在联想学习，但我们最远的动物表亲——径向对称的水母、海葵和珊瑚——却没有能够学习联想。*尽管光与电击有很多种组合，但海葵永远不会学会只对光做出反应而退缩。它们只会对电击本身做出退缩。在两侧对称动物中，联想学习无处不在，而在两侧对称动物之外，联想学习却明显缺失，这表明联想学习首先出现在早期两侧对称动物的大脑中。似乎在价态（将世界上的事物分为好与坏）出现的同时，利用经验改变被认为是好与坏的能力也出现了。
And yet, while associative learning is found across bilaterians, our most distant animal cousins—the radially symmetric jellyfish, anemones, and coral—are not capable of learning associations.* Despite many pairings of a light with an electric shock, an anemone will never learn to withdraw in response to just the light. They withdraw only from the shock itself. The ubiquitous presence of associative learning within Bilateria and the notable absence of it outside Bilateria suggests that associative learning first emerged in the brains of early bilaterians. It seems that at the same time valence—the categorizing of things in the world into good and bad—emerged, so too did the ability to use experience to change what is considered good and bad in the first place.
为什么像珊瑚和海葵这样的非对称动物，尽管经过六亿年的进化，它们还没有获得学习联想的能力吗？它们的生存策略根本不需要这种能力。
Why have non-bilaterian animals like coral and anemones, despite an additional six hundred million years of evolution, not acquired the ability to learn associations? Their survival strategy simply doesn’t require it.
具有联想学习能力的珊瑚虫的生存能力不会比没有联想学习能力的珊瑚虫好多少。珊瑚虫只是坐在原地，一动不动，等待食物游进它的触手。吞下任何触手触碰的东西并避开任何痛苦的事物这一硬编码策略在没有任何联想学习的情况下也能正常工作。相比之下，为转向而设计的大脑将面临独特的进化压力，需要根据经验调整其转向决策。能够记住避开之前在捕食者附近发现的化学物质的早期两侧对称动物的生存能力远高于无法记住的两侧对称动物。
A coral polyp with the ability of associative learning wouldn’t survive much better than one without associative learning. A coral polyp just sits in place, immobilized, waiting for food to swim into its tentacles. The hardcoded strategy of swallowing anything that touches its tentacles and withdrawing from anything painful works just fine, without any associative learning. In contrast, a brain designed for steering would have faced unique evolutionary pressure to adjust its steering decisions based on experience. An early bilaterian that could remember to avoid a chemical that had previously been found near predators would survive far better than a bilaterian that could not.
一旦动物开始接近特定事物并避开其他事物，调整好坏的能力就变成了生死攸关的问题。
Once animals began approaching specific things and avoiding others, the ability to tweak what was considered good and bad became a matter of life and death.
持续学习问题
The Continual Learning Problem
你的自动驾驶汽车不会随着你的驾驶而自动变得更好；你手机中的面部识别技术不会在你每次打开手机时自动变得更好。截至 2023 年，大多数现代人工智能系统都要经过一个训练过程，一旦训练完成，它们就会被送入现实世界供人们使用，但它们不再学习。这一直给人工智能系统带来一个问题——如果世界上的偶然事件以训练数据无法捕捉到的方式发生变化，那么这些人工智能系统就需要重新训练，否则它们会犯下灾难性的错误。如果新立法要求人们在道路左侧行驶，而人工智能系统被训练成只在道路右侧行驶，那么如果不经过明确的再训练，它们将无法灵活地适应新环境。
Your self-driving car doesn’t automatically get better as you drive; the facial-recognition technology in your phone doesn’t automatically get better each time you open your phone. As of 2023, most modern AI systems go through a process of training, and once trained, they are sent off into the world to be used, but they no longer learn. This has always presented a problem for AI systems—if the contingencies in the world change in a way not captured in the training data, then these AI systems need to be retrained, otherwise they will make catastrophic mistakes. If new legislation required people to drive on the left side of the road, and AI systems were trained to drive only on the right side of the road, they would not be able to flexibly adjust to the new environment without being explicitly retrained.
虽然现代人工智能系统的学习不是持续性的，但生物大脑的学习一直是持续性的。即使是我们的祖先线虫也别无选择，只能不断学习。事物之间的关联总是在变化。在某些环境中，食物上有盐；在其他环境中，没有食物的贫瘠岩石上也有盐。在某些环境中，食物在凉爽的温度下生长；在其他环境中，食物在温暖的温度下生长。在某些环境中，食物在明亮的地方；在其他环境中，捕食者在明亮的区域发现。最初的大脑需要一种机制，不仅要获得联想，还要快速改变这些联想，以适应不断变化的世界规则。巴甫洛夫是第一个发现这些古老机制的线索的人。
While learning in modern AI systems is not continual, learning in biological brains has always been continual. Even our ancestral nematode had no choice but to learn continually. The associations between things were always changing. In some environments, salt was found on food; in others, it was found on barren rocks without food. In some environments, food grew at cool temperatures; in others, it grew at warm temperatures. In some environments, food was found in bright areas; in others, predators were found in bright areas. The first brains needed a mechanism to not only acquire associations but also quickly change these associations to match the changing rules of the world. It was Pavlov who first found hints of these ancient mechanisms.
通过测量与食物配对的提示所引起的唾液分泌量，巴甫洛夫不仅能够观察到关联的存在，还能定量测量这些关联的强度——响应提示时分泌的唾液越多，关联越强。巴甫洛夫找到了一种测量记忆的方法。通过记录记忆随时间的变化，巴甫洛夫可以观察到持续学习的过程。
By measuring the quantity of saliva produced in response to cues that had been paired with food, Pavlov was able to not only observe the presence of associations, but also quantitatively measure the strength of these associations—the more saliva released in response to a cue, the stronger the association. Pavlov had found a way to measure memory. And by recording how memory changed over time, Pavlov could observe the process of continual learning.
事实上，巴甫洛夫条件反射中的联想总是随着每次新体验而增强或减弱。在巴甫洛夫的实验中，联想随着每次后续配对而增强——每次在喂食前蜂鸣声响起，狗在下次蜂鸣声响起时流的口水就越多。这个过程称为习得（联想被习得）。
Indeed, the associations in Pavlov’s conditional reflexes are always strengthening or weakening with each new experience. In Pavlov’s experiments, the associations strengthened with each subsequent pairing—each time the buzzer occurred before food was given, the more the dog salivated the next time the buzzer occurred. This process is called acquisition (the association was being acquired).
如果在学会这种关联之后，在没有食物的情况下展示蜂鸣器，那么这种关联的强度会随着每次尝试而减弱，这一过程被称为灭绝。
If after learning this association, the buzzer is presented in the absence of food, then the strength of the association fades with each trial, a process called extinction.
消退有两个有趣的特点。假设你打破了以前学到的联想——连续几次按蜂鸣器，但不给食物。正如预期的那样，狗最终会停止对蜂鸣器流口水。然而，如果你等几天，然后再按蜂鸣器，就会发生一件奇怪的事情：狗又开始对蜂鸣器做出反应而流口水。这被称为自发恢复：被打破的联想被迅速抑制，但实际上并没有被忘记；只要有足够的时间，它们就会重新出现。此外，如果在对被打破的联想（有蜂鸣器但没有食物）进行了长时间的试验后，你恢复了这种联想（再次按蜂鸣器并提供食物），那么旧的联想将比狗第一次体验到蜂鸣器和食物之间的联想时更快地重新学习。这被称为重新习得：旧的消失的联想比全新的联想重新习得的速度要快得多。
There are two interesting features of extinction. Suppose you break a previously learned association—sound the buzzer several times in a row, but don’t give food. As expected, dogs will eventually stop salivating at the buzzer. However, if you wait a few days and then sound the buzzer again, something odd happens: dogs start salivating in response to the buzzer again. This is called spontaneous recovery: broken associations are rapidly suppressed but not, in fact, unlearned; given enough time, they reemerge. Further, if after a long stretch of trials with a broken association (buzzer but no food), you reinstate the association (sound a buzzer and provide food again), the old association will be relearned far more rapidly than the first time the dog experienced the association between the buzzer and food. This is called reacquisition: old extinguished associations are reacquired faster than entirely new associations.
为什么联想会表现出自发的恢复和重新习得？让我们来思考一下联想学习进化的古代环境。假设一条虫子有过多次在盐旁边找到食物的经历。然后有一天，它闻到了盐，并朝盐走去，却没有找到食物。当这条虫子花了一个小时嗅探却没有找到食物后，这种联系就会消失，虫子开始朝其他线索走去，不再被盐吸引。如果两天后它再次闻到盐，它是朝盐走去还是远离盐更聪明？在这条虫子过去的所有经历中，除了最近的一次，当它闻到盐的时候，它也会找到食物。所以更明智的选择是再次朝盐走去——最近的一次经历可能是侥幸。这就是自发恢复的好处——它使一种原始形式的长期记忆能够在世界突发事件的短期变化中持续存在。当然，如果这条虫子接下来二十次闻到盐却找不到食物，这种联系最终可能会永久消失。
Why do associations show spontaneous recovery and reacquisition? Consider the ancient environment in which associative learning evolved. Suppose a worm has many experiences of finding food alongside salt. And then one day, it detects salt, steers toward it, and finds no food. After the worm spends an hour sniffing around without finding food, the association becomes extinguished, and the worm begins steering toward other cues, no longer attracted to salt. If two days later it detects salt again, would it be smarter to steer toward or away from it? In all of the worm’s past experiences, except the most recent one, when it smelled salt it also found food. And so the smarter choice would be to steer toward salt again—the most recent experience may have been a fluke. This is the benefit of spontaneous recovery—it enables a primitive form of long-term memory to persist through the tumult of short-term changes in the contingencies of the world. Of course, if the next twenty times the worm detects salt it fails to find food, the association may eventually be permanently extinguished.
图 4.2：联想学习的时间过程
Figure 4.2: The time course of associative learning
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
重新习得效应（加速重新学习以前被破坏的旧关联）在古代蠕虫中也因类似原因而进化。假设这条蠕虫在很久以前就已失去关联后，又在食物旁发现了盐。蠕虫应该多快重新加强盐和食物之间的关联？考虑到蠕虫拥有的长期记忆，快速重新学习这种关联是有意义的：在某些情况下，盐会导致食物，而现在似乎就是其中一种情况！因此，只要世界提供旧偶发性重新建立的暗示，旧关联就会重新出现。
The effect of reacqusition—the accelerated relearning of old previously broken associations—evolved in ancient worms for similar reasons. Suppose this same worm finds salt alongside food after the association was long ago extinguished. How quickly should the worm restrengthen the association between salt and food? It would make sense to relearn this association rapidly, given the long-term memory the worm has: In some cases, salt leads to food, and it seems that right now is one of those situations! Thus, old associations are primed to reemerge whenever the world provides hints that old contingencies are newly reestablished.
自发的恢复和重新获得使简单的操纵大脑能够驾驭不断变化的联想，暂时抑制当前不准确的旧联想，并记住并重新学习再次有效的断裂联想。
Spontaneous recovery and reacquisition enabled simple steering brains to navigate changing associations, temporarily suppress old associations that were currently inaccurate, and remember and relearn broken associations that became effective again.
第一批两侧对称动物利用这些习得、消退、自发恢复和重新习得的技巧来应对他们世界中不断变化的突发事件。这些持续学习的解决方案存在于许多动物的许多反射中，即使是像线虫这样最古老的动物，它们嵌入在最简单的神经回路中，从第一个大脑中继承而来，最初是为了在不断变化的世界中掌舵而制作的。古埃迪卡拉海。
The first bilaterians used these tricks of acquisition, extinction, spontaneous recovery, and reacquisition to navigate changing contingencies in their world. These solutions to continual learning are found across many reflexes in many animals, even the most ancient animals like the nematode, embedded in the simplest neural circuits, inherited from the first brain, originally crafted to make steering work in the ever-changing world of the ancient Ediacaran Sea.
信用分配问题
The Credit Assignment Problem
联想学习还带来了另一个问题：当动物获取食物时，事先不会有任何预测线索，而是会收到一系列线索。如果你将轻拍蛞蝓身体一侧的动作与电击配对，蛞蝓的大脑如何知道只将轻拍与电击联系起来，而不考虑当时存在的许多其他感官刺激，例如周围的温度、地面的质地或漂浮在海水中的各种化学物质？在机器学习中，这被称为信用分配问题：当某件事发生时，你会将预测该事的信用归功于先前的哪个线索？古老的两侧对称动物大脑只能进行最简单的学习，它采用了四种技巧来解决信用分配问题。这些技巧既粗糙又巧妙，它们成为了所有两侧对称动物后代神经元建立关联的基础机制。
Associative learning comes with another problem: When an animal gets food, there is never a single predictive cue beforehand but rather a whole swath of cues. If you pair a tap to the side of a slug with a shock, how does a slug’s brain know to associate only the tap with the shock and not the many other sensory stimuli that were present, such as the surrounding temperature, the texture of the ground, or the diverse chemicals floating around the seawater? In machine learning, this is called the credit assignment problem: When something happens, what previous cue do you give credit for predicting it? The ancient bilaterian brain, which was capable of only the simplest forms of learning, employed four tricks to solve the credit assignment problem. These tricks were both crude and clever, and they became foundational mechanisms for how neurons make associations in all their bilaterian descendants.
第一个技巧使用了所谓的“资格痕迹”。只有当敲击发生在电击前一秒时，蛞蝓才会将敲击与随后的电击联系起来。如果敲击发生在电击前两秒或更长时间，则不会产生任何关联。像敲击这样的刺激会产生短暂的资格痕迹，持续约一秒钟。只有在这个短暂的时间窗口内才能产生关联。这很聪明，因为它引用了一个合理的经验法则：对预测事情有用的刺激应该发生在你试图预测的事情之前。
The first trick used what are called eligibility traces. A slug will associate a tap with a subsequent shock only if the tap occurs one second before the shock. If the tap occurs two seconds or more before the shock, no association will be made. A stimulus like a tap creates a short eligibility trace that lasts for about a second. Only within this short time window can associations be made. This is clever, as it invokes a reasonable rule of thumb: stimuli that are useful for predicting things should occur right before the thing you are trying to predict.
第二个技巧是掩盖。当动物有多个预测线索可用时，它们的大脑倾向于选择最强的线索——强线索掩盖弱线索。如果在事件发生前同时出现明亮的光线和微弱的气味，则明亮的光线而不是微弱的气味将被用作预测线索。
The second trick was overshadowing. When animals have multiple predictive cues to use, their brains tend to pick the cues that are the strongest—strong cues overshadow weak cues. If a bright light and a weak odor are both present before an event, the bright light, not the weak odor, will be used as the predictive cue.
第三个技巧是潜在抑制——动物过去经常经历的刺激被抑制，无法产生未来的联想。换句话说，频繁的刺激被标记为不相关的背景噪音。潜在抑制是一种巧妙的方式，可以问“这次有什么不同？”如果一只蛞蝓已经体验过地面的当前质地和当前温度一千次，但从未体验过轻拍，那么这次轻拍更有可能被用作预测线索。
The third trick was latent inhibition—stimuli that animals regularly experienced in the past are inhibited from making future associations. In other words, frequent stimuli are flagged as irrelevant background noise. Latent inhibition is a clever way to ask, “What was different this time?” If a slug has experienced the current texture of the ground and the current temperature a thousand times but has never experienced a tap before, then the tap is far more likely to be used as a predictive cue.
第四个也是最后一个技巧信用分配问题就是阻断。一旦动物建立了预测线索和响应之间的关联，所有与预测线索重叠的进一步线索都会被阻止与该响应相关联。如果蛞蝓知道轻拍会导致电击，那么新的质地、温度或化学物质就会被阻止与电击相关联。阻断是一种坚持一个预测线索并避免冗余关联的方法。
The fourth and final trick for navigating the credit assignment problem was blocking. Once an animal has established an association between a predictive cue and a response, all further cues that overlap with the predictive cue are blocked from association with that response. If a slug has learned that a tap leads to shock, then a new texture, temperature, or chemical will be blocked from being associated with the shock. Blocking is a way to stick to one predictive cue and avoid redundant associations.
解决学分分配问题的四个技巧
The Original Four Tricks for Tackling the Credit Assignment Problem
资格追踪 ELIGIBILITY TRACES |
遮蔽 OVERSHADOWING |
潜伏性抑制 LATENT INHIBITION |
封锁 BLOCKING |
选择事件发生前0 到 1 秒之间发生的预测提示。 Pick the predictive cue that occurred between 0 to 1 second before the event. |
选择最强的预测线索。 Pick the predictive cue that was the strongest. |
选择您之前从未见过的预测线索。 Pick the predictive cue that you haven’t seen before. |
一旦获得预测线索，就坚持下去，并忽略其他线索。 Stick to predictive cues once you have them and ignore others. |
资格痕迹、遮蔽、潜在抑制和阻断在两侧对称动物中无处不在。巴甫洛夫在他的流涎狗的条件反射中发现了这些；它们也存在于人类的非自愿反射中；它们也出现在扁虫、线虫、蛞蝓、鱼、蜥蜴、鸟类、老鼠和几乎所有两侧对称动物的联想学习中在动物王国中。这些技巧可以帮助你完成学分分配这个问题早在最初的大脑进行联想学习的时候就已经出现了。
Eligibility traces, overshadowing, latent inhibition, and blocking are ubiquitous across Bilateria. Pavlov identified these in the conditional reflexes of his salivating dogs; they are found in the involuntary reflexes of humans; and they are seen in the associative learning of flatworms, nematodes, slugs, fish, lizards, birds, rats, and most every bilaterian in the animal kingdom. These tricks for navigating the credit assignment problem evolved as far back as the very first brains to make associative learning work.
这些技巧并非十全十美。在某些情况下，最佳预测线索可能出现在事件发生前一分钟，而不是一秒钟。在其他情况下，最佳预测线索可能是弱线索，而不是强线索。随着时间的推移，大脑进化出了更复杂的策略来解决信用分配问题（敬请期待突破 2和突破 3）。但第一种解决方案的残余——资格痕迹、遮蔽、潜在抑制和阻塞——仍然存在于现代大脑中。它们出现在我们的非自愿反射和最古老的大脑回路中。事实上，整个大脑被切除的老鼠，只剩下脊髓中的神经回路，仍然表现出潜在抑制，阻断和遮蔽。连同习得、消退、自发恢复和重新习得，这些技巧组合构成了联想学习的神经机制的基础，这些机制深深嵌入神经元、神经回路和大脑本身的内部运作中。
These tricks are hardly perfect. In some circumstances, the best predictive cue may occur one minute before the event, not one second before. In other circumstances, the best predictive cue may be the weak cue, not the strong one. Over time, brains evolved more sophisticated strategies for solving the credit assignment problem (stay tuned for breakthrough #2 and breakthrough #3). But the remnants of the first solutions—eligibility traces, overshadowing, latent inhibition, and blocking—still exist in modern brains. They are seen in our involuntary reflexes and in our most ancient brain circuits. Indeed, rats with their entire brains removed, with nothing left but their neural circuits in their spinal cord, still show latent inhibition, blocking, and overshadowing. Along with acquisition, extinction, spontaneous recovery, and reacquisition, this portfolio of tricks make up the foundation of the neural mechanisms of associative learning, mechanisms that are embedded deep into the inner workings of neurons, neural circuits, and brains themselves.
古老的学习机制
The Ancient Mechanisms of Learning
几千年来，两组哲学家一直在争论大脑和心灵之间的关系。一组是二元论者，如柏拉图、阿奎那和笛卡尔，他们认为心灵独立于大脑而存在。实体可能相互作用，但它们是不同的；心灵是超越物理的东西。唯物主义者，如卡纳达、德谟克利特、伊壁鸠鲁和霍布斯，认为无论心灵是什么，它都完全位于大脑的物理结构中。没有任何东西超越物理。这场争论仍在世界各地的哲学系中激烈进行。如果你读到这本书的这一部分，我会假设你倾向于唯物主义，你——和我一样——倾向于拒绝对事物的非物理解释，甚至是心灵。但是，通过站在唯物主义者一边，我们引入了几个问题，这些问题一开始很难用物理来解释，最明显的就是学习。
For thousands of years, two groups of philosophers have been debating the relationship between the brain and the mind. One group, the dualists, like Plato, Aquinas, and Descartes, argue that the mind exists separately from the brain. The entities might interact with each other, but they are distinct; the mind is something beyond the physical. The materialists, like Kanada, Democritus, Epicurus, and Hobbes, argued that whatever the mind is, it is located entirely in the physical structure of the brain. There is nothing beyond the physical. This debate still rages in philosophy departments around the world. If you have made it this far in the book, I will assume you lean on the side of materialism, that you—like me—tend to reject nonphysical explanations for things, even the mind. But by siding with the materialists, we introduce several issues that, at first, are hard to explain physically, the most obvious being learning.
你可以读一遍句子，然后立即大声重复。如果我们坚持唯物主义的观点，这意味着读这句话瞬间改变了大脑中的物理特性。任何导致学习的事情都会导致我们每个人大脑中 860 亿个神经元中的某些东西发生物理重组。跟踪对话、观看电影和学习系鞋带都必须改变我们大脑的物理特性。
You can read a sentence once and then immediately repeat it out loud. If we stick to a materialist view, this means that reading this sentence instantaneously changed something physical in your brain. Anything that leads to learning causes physical reorganization of something in the 86 billion neurons in each of our heads. Keeping track of a conversation, watching a movie, and learning to tie your shoes all must change the physicality of our brains.
数千年来，人们一直在推测学习的物理机制，甚至二元论者也对学习做出了唯物主义的解释。柏拉图认为，大脑就像一块蜡板，感知会在其中留下持久的印象；他认为记忆就是这些印象。笛卡尔认为，记忆是通过在大脑中创造新的“褶皱”而形成的，“就像这张纸被撕碎后留下的褶皱一样。”曾经被折叠过。”其他人推测记忆是持续的“振动”。这些想法都是错误的，尽管这并不是它们的发起者的错；当时，没有人理解神经系统的基本构造块，所以他们甚至无法开始设想学习如何发挥作用。
People have been speculating on the physical mechanisms of learning for thousands of years, and even the dualists pontificated on materialist explanations for learning. Plato believed that the brain was like a wax tablet in which perceptions left lasting impressions; he believed that memories were these impressions. Descartes argued that memories were formed through creating new “folds” in the brain, “not unlike the folds which remain in this paper after it has once been folded.” Others speculated that memories were persistent “vibrations.” These ideas were all wrong, although through no fault of their originators; at the time, no one understood even the basic building blocks of the nervous system, so they could not even begin to conceive of how learning might work.
20 世纪早期关于神经元的一系列发现为大脑提供了大量新的基础。神经元之间的连接（突触）的发现是最明显的新发现，它可能是大脑在学习过程中发生变化的因素。事实上，事实证明，学习不是源于印象、褶皱或振动，而是源于这些突触连接的变化。
The flurry of discoveries about neurons in the early twentieth century provided a host of new building blocks. The discovery of the connections between neurons—synapses—was the most obvious new thing that could presumably change in the brain during learning. Indeed, it turns out that learning rises not from impressions, folds, or vibrations but from changes to these synaptic connections.
当突触强度发生变化或形成新突触或移除旧突触时，就会发生学习。如果两个神经元之间的连接较弱，输入神经元将不得不发射许多脉冲才能使输出神经元产生脉冲。如果连接较强，输入神经元将不得不发射几个脉冲才能使输出神经元产生脉冲。输入神经元在脉冲响应时释放更多神经递质或突触后神经元增加蛋白质受体数量（因此对相同数量的神经递质反应更灵敏），突触的强度可以增加。
Learning occurs when synapses change their strength or when new synapses are formed or old synapses are removed. If the connection between two neurons is weak, the input neuron will have to fire many spikes to get the output neuron to spike. If the connection is strong, the input neuron will have to fire only a few spikes to get the output neuron to spike. Synapses can increase their strength by the input neuron releasing more neurotransmitter in response to a spike or the postsynaptic neuron increasing the number of protein receptors (hence more responsive to the same quantity of neurotransmitter).
突触有许多机制来选择何时加强或减弱。这些机制是极其古老的进化创新，源自第一批两侧对称动物的联想学习。例如，两侧对称动物神经元的突触中存在巧妙的蛋白质机制检测输入神经元是否在与输出神经元相似的时间窗口内激发。换句话说，单个连接可以检测输入（如由轻敲激活的感觉神经元）是否与输出（如由电击激活的运动神经元）同时激活。如果这些神经元同时被激活，这种蛋白质机制就会触发一个强化突触的过程。*因此，下次轻敲神经元激活时，它会自行激活运动神经元（因为神经元之间的连接得到了加强），现在你就有了条件反射。这种学习机制被称为赫布学习，以心理学家唐纳德·赫布的名字命名，他在 20 世纪 40 年代雄辩地假设了这种机制的存在，比这种机制被发现早了几十年。赫布学习通常被称为“一起激发的神经元会连接在一起”的规则。
Synapses have many mechanisms for choosing when to strengthen or weaken. These mechanisms are extremely old evolutionary innovations, originating from the associative learning of the first bilaterians. For example, there is clever protein machinery in the synapses of bilaterian neurons that detect whether the input neuron fired within a similar time window as the output neuron. In other words, individual connections can detect whether an input (like the sensory neuron activated by a tap) is activated at the same time as an output (like the motor neuron activated by a shock). And if these neurons are activated at the same time, this protein machinery triggers a process that strengthens the synapse.* Thus, next time the tap neuron activates, it activates the motor neuron on its own (because the connection between the neurons was strengthened), and you now have a conditional reflex. This learning mechanism is called Hebbian learning after the psychologist Donald Hebb, who eloquently hypothesized the existence of such a mechanism in the 1940s, several decades before the mechanism was discovered. Hebbian learning is often referred to as the rule that “neurons that fire together wire together.”
图 4.3
Figure 4.3
图为 Max Bennett
Figure by Max Bennett
但改变突触强度的逻辑比这更复杂。突触中有测量时间的分子机制，只有当输入神经元恰好在输出神经元之前激发时，才会建立关联，从而实现资格追踪的技巧。神经调节剂如血清素和多巴胺可以改变突触的学习规则；一些突触只有在多巴胺或血清素受体也被激活时才会进行赫布学习，从而使神经调节剂能够控制突触建立新关联的能力。如果蠕虫嗅到化学物质然后找到食物，它的大脑就会充满多巴胺，然后可以触发特定突触的强化。
But the logic of changing synaptic strengths gets more complex than this. There are molecular mechanisms in synapses to measure timing, whereby associations are built only if the input neuron fires right before the output neuron, thereby enabling the trick of eligibility traces. Neuromodulators like serotonin and dopamine can modify the learning rules of synapses; some synapses undergo Hebbian learning only when dopamine or serotonin receptors are also activated, thereby enabling neuromodulators to gate the ability of synapses to build new associations. A worm that sniffs a chemical and then finds food has its brain flooded with dopamine, which then can trigger the strengthening of specific synapses.
虽然我们还没有完全了解神经元自我重组的所有机制，但这些机制非常相似在两侧对称动物中，线虫大脑中的神经元改变突触的方式与大脑中的神经元大致相同。相比之下，当我们检查非两侧对称动物（如珊瑚虫）的神经元和突触时，我们找不到相同的机制；例如，它们缺乏某些已知的蛋白质参与赫布学习。但考虑到我们的进化历史，这是可以预料的：如果我们与珊瑚虫共同的祖先没有联想学习，那么我们应该预料到它们缺乏这种学习所依赖的机制。
Although we don’t yet fully understand all the mechanisms by which neurons rewire themselves, these mechanisms are remarkably similar among bilaterians; the neurons in the brain of a nematode change their synapses in largely the same way as the neurons in your brain. In contrast, when we examine the neurons and synapses of non-bilaterians like coral polyps, we do not find the same machinery; for example, they lack certain proteins known to be involved in Hebbian learning. But given our evolutionary past, this is to be expected: if our shared ancestor with coral polyps had no associative learning, then we should expect them to lack the mechanisms that underly such learning.
学习的起点很低。虽然早期的两侧对称动物是第一个学会联想的动物，但它们仍然无法学习大多数事物。它们无法学会将相隔超过几秒钟的事件联系起来；它们无法学会预测事物的准确时间；它们无法学会识别物体；它们无法识别世界上的模式；它们无法学会识别位置或方向。
Learning had humble beginnings. While early bilaterians were the first to learn associations, they were still unable to learn most things. They could not learn to associate events separated by more than a few seconds; they could not learn to predict the exact timing of things; they could not learn to recognize objects; they could not recognize patterns in the world; and they could not learn to recognize locations or directions.
但是，人类大脑重新连接自身、将事物联系起来的能力并不是人类独有的超能力，而是我们从生活在 5.5 亿年前的古代两侧对称动物祖先那里继承而来。之后的所有学习壮举（学习空间地图、语言、物体识别、音乐和其他一切的能力）都是建立在这些相同的学习机制之上的。从两侧对称动物的大脑开始，学习的进化主要是寻找现有突触学习机制的新应用的过程，而不会改变学习机制本身。
But still, the ability of the human brain to rewire itself, to make associations between things, is not a uniquely human superpower but one we inherited from this ancient bilaterian ancestor that lived over 550 million years ago. All the feats of learning that followed (the ability to learn spatial maps, language, object recognition, music, and everything else) were built on these same learning mechanisms. From the bilaterian brain onward, the evolution of learning was primarily a process of finding new applications of preexisting synaptic learning mechanisms, without changing the learning mechanisms themselves.
学习并不是第一个大脑的核心功能；它只是一个功能，一个优化决策的技巧。联想、预测和学习的出现是为了调整事物的善恶。从某种意义上说，接下来的进化故事就是学习从大脑的一个可爱功能转变为其核心功能的故事。事实上，大脑进化的下一个突破是关于一种全新的学习形式，这种形式之所以可能，只是因为它建立在价态、情感和联想学习的基础上。
Learning was not the core function of the first brain; it was merely a feature, a trick to optimize steering decisions. Association, prediction, and learning emerged for tweaking the goodness and badness of things. In some sense, the evolutionary story that will follow is one of learning being transformed from a cute feature of the brain to its core function. Indeed, the next breakthrough in brain evolution was all about a brilliant new form of learning, one that was possible only because it was built on the foundation of valence, affect, and associative learning.
突破 1：转向
Summary of Breakthrough #1: Steering
大约 5.5 亿年前，我们的祖先从一种径向对称的无脑动物（如珊瑚虫）转变为一种双侧对称的有脑动物（如线虫）。虽然这一转变过程中发生了许多神经系统变化，但其中令人惊讶的广泛变化可以通过实现一项独特突破的视角来理解：即通过转向导航。这些包括：
Our ancestors from around 550 million years ago transitioned from a radially symmetric brainless animal, like a coral polyp, to a bilaterally symmetric brain-enabled animal, like a nematode. And while many neurological changes occurred across this transition, a surprisingly broad set of them can be understood through the lens of enabling a singular breakthrough: that of navigating by steering. These include:
- 双侧身体计划将导航选择减少到两个简单的选项：前进或转弯
- A bilateral body plan that reduces navigational choices to two simple options: go forward or turn
- 一种价态的神经结构，其中刺激在进化过程中被硬编码为好与坏
- A neural architecture for valence in which stimuli is evolutionarily hard-coded into good and bad
- 根据内部状态调节价态反应的机制
- Mechanisms for modulating valence responses based on internal states
- 不同价态的神经元可以整合到一个单一的转向决策电路中（因此我们将一大群神经元称为大脑）
- Circuits whereby different valence neurons can be integrated into a singular steering decision (hence a big cluster of neurons we identify as a brain)
- 做出离开或留下的持续决定的情感状态
- Affective states for making persistent decisions as to whether to leave or stay
- 面临困难时运动能量管理的压力反应
- The stress response for energy management of movements in the presence of hardship
- 根据以往经验改变转向决策的联想学习
- Associative learning for changing steering decisions based on previous experience
- 自发恢复和重新获得，以应对世界上不断变化的突发事件（即使不完美，也能使持续学习发挥作用）
- Spontaneous recovery and reacquisition for dealing with changing contingencies in the world (making continual learning work, even if imperfectly)
- 资格追踪、遮蔽、潜在抑制和阻塞，用于（不完美地）解决信用分配问题
- Eligibility traces, overshadowing, latent inhibition, and blocking for (imperfectly) tackling the credit assignment problem
所有这些变化使得转向成为可能，并巩固了我们的祖先作为第一批通过导航生存下来的大型多细胞动物的地位——不是用微观细胞螺旋桨，而是用肌肉和神经元移动。所有这些变化，以及它们所催生的掠食生态系统，为第二次突破奠定了基础，也就是学习终于在我们大脑的功能中发挥了核心作用。
All of these changes made steering possible and solidified our ancestors’ place as the first large multicellular animals who survived by navigating—moving not with microscopic cellular propellers but with muscles and neurons. And all these changes, along with the predatory ecosystem they begot, laid the foundation for breakthrough #2, which was when learning finally took its central role in the function of our brains.
突破＃2
Breakthrough #2
强化和第一批脊椎动物
Reinforcing and the First Vertebrates
5亿年前的大脑
Your brain 500 million years ago
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
5
5
寒武纪生命大爆发
The Cambrian Explosion
要到达大脑进化的下一个里程碑，我们必须离开第一批两侧对称动物扭动身体的时代，向前跳五千万年。这将带我们进入古代世界，即寒武纪，一个从 5.4 亿年前到 4.85 亿年前的时代。
TO GET TO the next milestone in brain evolution, we must leave the era when the first bilaterians were wiggling around and jump forward fifty million years. The ancient world this brings us to is the Cambrian period, an era that stretched from 540 to 485 million years ago.
如果你环视寒武纪，你会看到一个与更古老的埃迪卡拉纪截然不同的世界。埃迪卡拉纪使海底变成绿色的粘稠微生物垫早已消退，取而代之的是人们更为熟悉的沙质海底。埃迪卡拉纪那些有理智、行动缓慢、体型娇小的生物将被一个熙熙攘攘的大型移动动物动物园所取代，这些动物形态和大小各异。这不像是你喜欢的动物园——这是一个由节肢动物统治的世界，节肢动物是昆虫、蜘蛛和甲壳类动物的祖先。这些节肢动物比它们的现代后代可怕得多；它们体型巨大，长着令人毛骨悚然的超大爪子和装甲壳。有些甚至长达五英尺多。
If you peered around the Cambrian, you would see a world very different from the older Ediacaran. The gooey microbial mats of the Ediacaran that turned the ocean floor green would have long since faded and given way to a more familiar sandy underbelly. The sensile, slow, and small creatures of the Ediacaran would have been replaced by a bustling zoo of large mobile animals as varied in form as in size. This wouldn’t resemble a zoo you would enjoy—this was a world ruled by arthropods, the ancestors of insects, spiders, and crustaceans. These arthropods were far more terrifying than their modern descendants; they were massive and armed with hauntingly oversize claws and armored shells. Some grew to over five feet long.
在我们类似线虫的祖先身上发现转向能力加速了捕食的进化军备竞赛。这引发了现在所谓的寒武纪生命大爆发，这是地球上有史以来动物多样性最显著的扩张。埃迪卡拉纪化石稀有且备受追捧，但如果你挖得足够深，寒武纪化石随处可见，它们包含了令人难以置信的多样性。在埃迪卡拉纪，有脑的动物是海底的卑微居民，比它们的无脑动物近亲如珊瑚和海葵体型更小、数量更少。然而，在寒武纪，有脑的动物开始统治动物界。
The discovery of steering in our nematode-like ancestor accelerated the evolutionary arms race of predation. This triggered what is now known as the Cambrian explosion, the most dramatic expansion in the diversity of animal life Earth has ever seen. Ediacaran fossils are rare and sought after, but Cambrian fossils, if you dig deep enough, are all over the place, and they encompass a mind-boggling diversity of creatures. During the Ediacaran period, animals with brains were humble inhabitants of the seafloor, smaller and less numerous than their brainless animal cousins like the coral and anemones. During the Cambrian period, however, animals with brains began their reign over the animal kingdom.
图 5.1：寒武纪世界
Figure 5.1: The Cambrian world
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
蠕虫奶奶的一个谱系基本保持不变，体型缩小，成为今天的线虫。另一个谱系成为这个时代的主人，即节肢动物。这些节肢动物的谱系会独立发展出自己的大脑结构和智力。有些节肢动物，如蚂蚁和蜜蜂，会变得非常聪明。但节肢动物和线虫谱系都不是我们的。在寒武纪的可怕生物中，我们的祖先可能并不显眼；它们比早期的两侧对称动物大不了多少，只有几英寸长，数量也不是特别多。但如果你发现它们，它们看起来会让人耳目一新——它们很像现代鱼。
One lineage of Grandma Worm remained relatively unchanged and shrank in size, becoming the nematodes of today. Another lineage became the masters of this era, the arthropods. Lineages of these arthropods would independently develop their own brain structures with their own intellectual abilities. Some, such as the ants and honeybees, would go on to become impressively smart. But neither the arthropod nor the nematode lineage is ours. Our ancestors were likely not very conspicuous in the Cambrian cacophony of terrifying creatures; they were barely bigger than early bilaterians, only a few inches long, and not particularly numerous. But if you spotted them, they would have looked refreshingly familiar—they would have resembled a modern fish.
这些古鱼的化石记录显示出几个熟悉的特征。它们有鳍、鳃、脊髓、两只眼睛、鼻孔和一颗心脏。这些生物化石中最容易发现的特征是脊柱，即包裹和保护脊髓的厚而相互交错的骨头。事实上，分类学家将这种古代鱼类祖先的后代称为脊椎动物。但在这些早期脊椎动物中出现的所有熟悉的变化中，最引人注目的无疑是大脑。
Fossil records of these ancient fish show several familiar features. They had fins, gills, a spinal cord, two eyes, nostrils, and a heart. The easiest-to-spot feature in fossils of these creatures is the vertebral column, the thick interlocking bones that encased and protected their spinal cord. Indeed, taxonomists refer to the descendants of this ancient fishlike ancestor as vertebrates. But of all the familiar changes that emerged in these early vertebrates, the most remarkable was surely the brain.
脊椎动物大脑模板
The Vertebrate Brain Template
无脊椎动物（线虫、蚂蚁、蜜蜂、蚯蚓）的大脑与人类的大脑没有明显相似的结构。进化人类与无脊椎动物之间的距离太远了；我们的大脑源自我们两侧对称动物祖先的一个太基本的模板，因此无法揭示任何共同的结构。但是，当我们观察即使是最遥远的脊椎动物的大脑时，比如无颌七鳃鳗——我们最近的共同祖先是五亿多年前的第一个脊椎动物——我们会看到一个大脑不仅共享一些相同的结构，而且共享大部分结构。
The brains of invertebrates (nematodes, ants, bees, earthworms) have no recognizably similar structures to the brains of humans. The evolutionary distance between humans and invertebrates is too distant; our brains are derived from too basic a template in our bilaterian ancestor to reveal any common structures. But when we peer into the brain of even the most distant vertebrates, such as the jawless lamprey fish—with whom our most recent common ancestor was the first vertebrate over five hundred million years ago—we see a brain that shares not only some of the same structures but most of them.
图 5.2：我们的寒武纪祖先
Figure 5.2: Our Cambrian ancestors
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
寒武纪生命大爆发的高温锻造出了脊椎动物的大脑模板，直到今天，这些早期鱼类生物的所有后代都共享着这种模板。如果你想速成人类大脑的工作原理，那么了解鱼脑的工作原理就等于成功了一半。
From the heat of the Cambrian explosion was forged the vertebrate brain template, one that, even today, is shared across all the descendants of these early fishlike creatures. If you want a crash course in how the human brain works, learning how the fish brain works will get you half of the way there.
从鱼类到人类，所有脊椎动物胚胎的大脑都以相同的初始步骤发育。首先，大脑分化为三个球状结构，构成支撑所有脊椎动物大脑的三个主要结构：前脑、中脑和后脑。其次，前脑展开为两个子系统。其中一个子系统继续发展成为皮层和基底神经节，另一个子系统继续发展成为丘脑和下丘脑。
The brains of all vertebrate embryos, from fish to humans, develop in the same initial steps. First, brains differentiate into three bulbs, making up the three primary structures that scaffold all vertebrate brains: a forebrain, midbrain, and hindbrain. Second, the forebrain unfolds into two subsystems. One of these goes on to become the cortex and the basal ganglia, and the other goes on to become the thalamus and the hypothalamus.
这导致了所有脊椎动物的大脑都有六个主要结构：皮层、基底神经节、丘脑、下丘脑、中脑和后脑。这些结构揭示了它们的共同祖先，在现代脊椎动物中非常相似（皮层除外，在某些脊椎动物中，如哺乳动物，皮层具有独特的修饰；敬请期待突破#3）。人类基底神经节、丘脑、下丘脑、中脑和后脑的回路与鱼类的回路极其相似。
This results in the six main structures found in all vertebrate brains: the cortex, basal ganglia, thalamus, hypothalamus, midbrain, and hindbrain. Revealing their common ancestry, these structures are remarkably similar across modern vertebrates (except for the cortex, which has unique modifications in some vertebrates, such as mammals; stay tuned for breakthrough #3). The circuitry of the human basal ganglia, thalamus, hypothalamus, midbrain, and hindbrain and that of a fish are incredibly similar.
图 5.3：脊椎动物的共同胚胎发育
Figure 5.3: The shared embryonic development of vertebrates
原创艺术由 Mesa Schumacher 创作
Original art by Mesa Schumacher
第一批动物赋予了我们神经元。然后早期的两侧对称动物赋予了我们大脑，将这些神经元聚集到中央电路中，连接起第一个效价、情感和关联的系统。但正是早期的脊椎动物将早期两侧对称动物的这种简单的原始大脑改造成一个真正的机器，一个具有亚基、层和处理系统的机器。
The first animals gifted us neurons. Then early bilaterians gifted us brains, clustering these neurons into centralized circuits, wiring up the first system for valence, affect, and association. But it was early vertebrates who transformed this simple proto-brain of early bilaterians into a true machine, one with subunits, layers, and processing systems.
当然，问题是早期脊椎动物的大脑有什么功能？
The question is, of course, what did this early vertebrate brain do?
桑代克的鸡
Thorndike’s Chickens
就在俄罗斯的伊万·巴甫洛夫揭示条件反射的内部运作机制的同时，一位名叫爱德华·桑代克的美国心理学家正在从不同的角度探索动物的学习。
Around the same time that Ivan Pavlov was unraveling the inner workings of conditional reflexes in Russia, an American psychologist by the name of Edward Thorndike was probing animal learning from a different perspective.
图 5.4：第一批脊椎动物的大脑
Figure 5.4: The brain of the first vertebrates
原创艺术由 Mesa Schumacher 创作
Original art by Mesa Schumacher
1896 年，爱德华·桑代克发现自己身处一个满是鸡的房间。桑代克当时刚刚入读哈佛大学的心理学硕士课程。他的主要兴趣是研究儿童的学习方式：我们如何才能最好地教孩子新东西？他有很多实验想法，但令桑代克懊恼的是，哈佛大学不允许他在人类儿童身上进行实验。因此，桑代克别无选择，只能专注于更容易获得的对象：鸡、猫和狗。
In 1896, Edward Thorndike found himself in a room full of chickens. Thorndike had recently enrolled in Harvard’s master’s program in psychology. His main interest was studying how children learn: How best can we teach children new things? He had numerous ideas for experiments, but to Thorndike’s chagrin, Harvard would not allow him to conduct experiments on human children. So Thorndike had no choice but to focus on subjects that were easier to obtain: chickens, cats, and dogs.
对桑代克来说，这并非全是坏事。作为一名坚定的达尔文主义者，他坚信鸡、猫、狗和人类的学习应该有共同的原则。如果这些动物有共同的祖先，那么它们都应该继承了类似的学习机制。通过探究这些其他动物的学习方式，他相信自己也许能够阐明人类学习的原理。
To Thorndike, this wasn’t all bad. A staunch Darwinist, he was unwavering in his view that there should be common principles in the learning of chickens, cats, dogs, and humans. If these animals shared a common ancestor, then they all should have inherited similar learning mechanisms. By probing how these other animals learned, he believed he might be able to also illuminate the principles of how humans learned.
桑代克非常害羞，但又非常聪明，所以他也许是从事他开创的孤独、一丝不苟、无可否认聪明的动物研究的最佳人选。巴甫洛夫在中年时完成了开创性的心理学研究，当时他已经是一位声名显赫的生理学家，但桑代克最著名的研究成果却是他的第一篇。他最著名的是他 1898 年发表的博士论文，当时他 23 岁。他的论文是《动物智力：动物联想过程的实验研究》。
Thorndike was both extremely shy and incredibly smart, so he was perhaps the perfect person to engage in the solitary, meticulously repetitive, and undeniably clever animal studies that he pioneered. Pavlov did his groundbreaking psychology work when he was middle-aged, after an already famed career as a physiologist, but Thorndike’s most famous work was his first. It was his doctoral dissertation, published in 1898, when he was twenty-three, for which he is most well known. His dissertation: “Animal Intelligence: An Experimental Study of the Associative Processes in Animals.”
桑代克的天才之处，就像巴甫洛夫一样，在于他如何无望地减少将复杂的理论问题转化为简单的可测量实验。巴甫洛夫通过测量蜂鸣器响起时唾液分泌量来探索学习。桑代克通过测量动物学会逃离他所谓的“谜题盒”的速度来探索学习。
Thorndike’s genius, like Pavlov’s, was in how he reduced hopelessly complex theoretical problems to simple measurable experiments. Pavlov explored learning by measuring the amount of saliva released in response to a buzzer. Thorndike explored learning by measuring the speed with which animals learned to escape from what he called puzzle boxes.
桑代克建造了许多笼子，每个笼子里都有不同的谜题，如果正确解决，就会打开一扇逃生门。这些谜题并不特别复杂——有些有门闩，推开门就可以打开；有些有隐藏的按钮；有些有需要拉动的环。有时解谜不需要物理装置，桑代克只需在动物做出特定动作（例如舔自己）时手动打开门即可。他将各种动物放在这些笼子里，在外面放上食物以激励动物从箱子里出来，并精确测量它们解开谜题所需的时间。
Thorndike constructed a multitude of cages, each with a different puzzle inside that, if solved correctly, would open an escape door. These puzzles weren’t particularly complex—some had latches that when pushed would open the door; others had hidden buttons; others had hoops to pull. Sometimes the puzzle did not require a physical contraption, and Thorndike would just manually open the door whenever the animal did something specific, such as lick itself. He placed various animals in these cages, put food outside to motivate the animals to get out of the boxes, and measured exactly how long it took them to figure out the puzzle.
一旦动物逃脱，他就会记录动物的时间，然后让动物一遍又一遍地重复。他会计算动物在第一次尝试中解决给定难题的平均时间，将其与第二次尝试的时间进行比较，并计算出它们在多达一百次尝试后解决难题的速度。
Once the animal escaped, he would record the animal’s time, and then have the animal do it again, and again, and again. He would calculate the average time it took animals to solve a given puzzle on their first trial, compare that with the time for their second, and go all the way to how fast they solved it after as many as a hundred trials.
桑代克最初想探究模仿的动态，他认为这是学习的一个特征，存在于许多动物物种。他让未经训练的猫观看受过训练的猫从各种谜题盒中逃脱，看看这是否对它们自己的学习有任何影响。换句话说，猫能通过模仿来学习吗？当时答案似乎是否定的；它们通过观察并没有变得更好（请注意，有些动物可以做到这一点；敬请期待突破#4）。但在这次失败中，他发现了一些令人惊讶的事情。他发现这些动物确实都具有一种共同的学习机制——只是不是他最初预期的那种。
Thorndike originally wanted to probe the dynamics of imitation, a feature of learning he believed would exist across many animal species. He allowed untrained cats to watch trained cats escape from various puzzle boxes to see if it had any effect on their own learning. In other words, could cats learn through imitation? It seemed at the time that the answer was no; they didn’t get any better by watching (note that some animals can do this; stay tuned for breakthrough #4). But in this failure, he discovered something surprising. He found that these animals did all share a learning mechanism—it just wasn’t the one he originally expected.
图 5.6：动物通过反复试验进行学习
Figure 5.6: Animals learning through trial and error
桑代克图片，1898 年
Images from Thorndike, 1898
刚被关进笼子里时，猫会尝试各种各样的行为：抓栅栏、推天花板、挖门、嚎叫、试图挤过栅栏、在笼子里走来走去。最后，猫会不小心按到按钮或拉到环，门就会打开；猫会出来，开心地吃掉它的奖品。动物们重复让它们离开盒子的行为的速度越来越快。经过多次试验后，猫会停止任何原来的行为，并立即执行逃跑所需的动作。这些猫正在通过反复试验进行学习。他可以通过动物逃脱所需时间的逐渐减少来量化这种反复试验的学习（图 5.6）。
When first placed in a cage, the cat would try a whole host of behaviors: scratching at the bars, pushing at the ceiling, digging at the door, howling, trying to squeeze through the bars, pacing around the cage. Eventually the cat would accidently press the button or pull the hoop, and the door would open; the cat would exit and happily eat its prize. The animals became progressively faster at repeating the behaviors that got them out of the box. After many trials, cats stopped doing any of their original behaviors and immediately performed the actions required to escape. These cats were learning through trial and error. He could quantify this trial-and-error learning with the gradual decay in the time it took for animals to escape (f igure 5.6).
最令人惊讶的是，如此多的智能行为都是从简单的反复试验中产生的。经过足够多的试验，这些动物可以毫不费力地完成一系列极其复杂的动作。最初人们认为，解释动物这种智能行为的唯一方法是通过某种洞察力、模仿或计划的概念，但桑代克表明，简单的反复试验就是动物真正需要的。桑代克将他的研究结果总结在他现在著名的效果定律中：
What was most surprising was how much intelligent behavior emerged from something as simple as trial-and-error learning. After enough trials, these animals could effortlessly perform incredibly complex sequences of actions. It was originally believed that the only way to explain such intelligent behavior in animals was through some notion of insight or imitation or planning, but Thorndike showed how simple trial and error was all an animal really needed. Thorndike summarized his result in his now famous law of effect:
在特定情况下产生令人满意的效果的反应更有可能发生再次出现这种情况，并做出回应产生令人不适影响的事件在那种情况下不太可能再次发生。
Responses that produce a satisfying effect in a particular situation become more likely to occur again in that situation, and responses that produce a discomforting effect become less likely to occur again in that situation.
动物通过首先执行随机探索性行为，然后根据效价结果调整未来行为来学习——正效价会强化最近执行的行为，而负效价会取消强化之前执行的行为。在桑代克最初的研究之后的几十年里，令人满意和不舒服这两个术语逐渐失宠；它们对实际的内部感觉或感受有令人不舒服的暗示。包括桑代克在内的心理学家最终用强化和惩罚取代了令人满意和不舒服这两个术语。
Animals learn by first performing random exploratory actions and then adjusting future actions based on valence outcomes—positive valence reinforces recently performed actions, and negative valence un-reinforces previously performed actions. The terms satisfying and discomforting went out of favor over the decades following Thorndike’s original research; they had an uncomfortable allusion to an actual internal sensation or feeling. Psychologists, including Thorndike, eventually replaced the terms satisfying and discomforting with reinforcing and punishing.
桑代克的思想继承者之一 BF 斯金纳甚至认为，所有动物行为，甚至人类行为，都是反复试验的结果。正如我们将在本书的第 3、4和5个突破中看到的那样，BF 斯金纳被证明是错误的。但是，虽然反复试验并不能解释动物学习的所有过程，但它却支撑了其中很大一部分。
One of Thorndike’s intellectual successors, B. F. Skinner, went so far as to suggest that all animal behavior, even in humans, was a consequence of nothing more than trial and error. As we will see with breakthroughs #3, #4, and #5 in this book, B. F. Skinner turned out to be wrong. But while trial and error does not explain all of animal learning, it undergirds a surprisingly large portion of it.
桑代克最初的研究对象是猫、狗和鸟，这些动物在大约 3.5 亿年前拥有共同的祖先。但那些更远的脊椎动物近亲，也就是我们在 5 亿年前拥有共同祖先的动物呢？它们也是通过反复试验来学习的吗？
Thorndike’s original research was on cats, dogs, and birds—animals that share a common ancestor around 350 million years ago. But what about more distant vertebrate cousins, those that we share an ancestor with as far back as 500 million years ago? Do they too learn through trial and error?
1898 年论文发表一年后，桑代克发表了另一篇论文，展示了对不同动物：鱼进行的相同研究的结果。
A year after his 1898 dissertation, Thorndike published an additional note showing the results of these same studies performed on a different animal: fish.
鱼类的惊人智慧
The Surprising Smarts of Fish
如果说脊椎动物中有什么成员最受人类的偏见，那一定是鱼类。鱼类是愚蠢的这一观念根植于许多文化中。我们都听说过这样一个民间传说：鱼类的记忆力不能超过三秒钟。也许所有这些偏见都是可以预料到的；鱼类是与我们最不相似的脊椎动物。但这种偏见是没有根据的；鱼类比我们认为的要聪明得多。
If there is any member of the vertebrate group that humans bear the most prejudice against, it is fish. The idea that fish are, well, dumb is embedded in many cultures. We have all heard the folklore that fish cannot retain memories for more than three seconds. Perhaps all this prejudice is to be expected; fish are the vertebrates that are the least like us. But this prejudice is unfounded; fish are far smarter than we give them credit for.
在桑代克最初的实验中，他将一条鱼放在一个鱼缸里，鱼缸有一系列透明的墙壁，墙壁上都有隐藏的开口。他将鱼放在鱼缸的一侧（明亮的光线下，鱼不喜欢），鱼缸的另一侧是一个理想的位置（黑暗，鱼喜欢）。起初，鱼尝试了许多随机的东西穿过鱼缸，经常撞到透明墙壁的某些部分。最后，鱼找到了一个缝隙，穿过了下一个墙壁。然后它重复这个过程，直到找到下一个缝隙。一旦鱼穿过了所有的墙壁到达另一边，桑代克就会把它捡起来，把它带回起点，让它重新开始，每次都记录鱼到达另一边所花的时间。就像桑代克的猫通过反复试验学会了逃离拼图盒一样，他的鱼也学会了快速穿过每个隐藏的开口，逃离鱼缸明亮的一面。
In Thorndike’s original experiment, he put a fish in a tank with a series of transparent walls with hidden openings. He put the fish on one side of the tank (in a bright light, which fish dislike), and on the other side of the tank was a desirable location (the dark, which fish prefer). At first, the fish tried lots of random things to get across the tank, frequently banging into parts of the transparent wall. Eventually the fish found one of the gaps and made it through to the next wall. It then repeated the process until it found the next gap. Once the fish made it past all the walls to the other side, Thorndike picked it up, brought it back to the beginning, and had it start again, each time clocking how long it took the fish to get to the other side. Just as Thorndike’s cats learned to escape puzzle boxes through trial and error, so did his fish learn to quickly zip through each of the hidden openings to escape the bright side of the tank.
鱼类通过反复试验来学习任意动作序列的能力已被多次复制。鱼类可以学会寻找并推动特定按钮来获取食物；鱼可以学会通过一个小的逃生舱来避免被网捕获；鱼甚至可以学会跳过铁环去获取食物。经过训练后，鱼可以记住如何完成这些任务几个月甚至几年。在所有这些测试中，学习的过程都是相同的：鱼尝试一些相对随机的动作，然后根据得到强化的内容逐步改进它们的行为。事实上，桑代克的试错学习通常有另一个名字：强化学习。
This ability of fish to learn arbitrary sequences of actions through trial and error has been replicated many times. Fish can learn to find and push a specific button to get food; fish can learn to swim through a small escape hatch to avoid getting caught in a net; and fish can even learn to jump through hoops to get food. Fish can remember how to do these tasks for months or even years after being trained. The process of learning is the same in all these tests: fish try some relatively random actions and then progressively refine their behavior depending on what gets reinforced. Indeed, Thorndike’s trial and error learning often goes by another name: reinforcement learning.
如果你试图教线虫、扁虫或蛞蝓等简单的两侧对称动物执行任何这些任务，它都会失败。线虫无法被训练执行任意的动作序列；它永远不会学会如何穿过铁环去获取食物。
If you tried to teach a simple bilaterian like a nematode, flatworm, or slug to perform any of these tasks, it would fail. A nematode cannot be trained to perform arbitrary sequences of actions; it will never learn to navigate through hoops to get food.
在接下来的四章中，我们将探索强化学习的挑战，并了解为什么祖先的两侧对称动物（如现代线虫）无法以这种方式学习。我们将了解第一个脊椎动物的大脑是如何工作的，它们是如何克服这些早期的挑战的，以及这些大脑是如何发展成为通用强化学习机器的。
Over the next four chapters we will explore the challenges of reinforcement learning and learn why ancestral bilaterians, like modern nematodes, were unable to learn this way. We will learn about how the first vertebrate brains worked, how they overcame these earlier challenges, and how these brains flowered into general reinforcement learning machines.
第二个突破是强化学习：通过反复试验来学习任意动作序列的能力。桑代克的反复试验学习理念听起来很简单——强化行为，导致好事的行为，惩罚导致坏事的行为。但这是一个例子，表明我们对智力上什么是容易的、什么是困难的直觉是错误的。直到科学家试图让人工智能系统通过强化来学习时，他们才意识到这并不像桑代克想象的那么容易。
The second breakthrough was reinforcement learning: the ability to learn arbitrary sequences of actions through trial and error. Thorndike’s idea of trial-and-error learning sounds so simple—reinforce behaviors that lead to good things and punish behaviors that lead to bad things. But this is an example where our intuitions about what is intellectually easy and what is hard are mistaken. It was only when scientists tried to get AI systems to learn through reinforcement that they realized that it wasn’t as easy as Thorndike had thought.
6
6
时间差分学习的演变
The Evolution of Temporal Difference Learning
第一个强化学习计算机算法是由普林斯顿大学的博士生马文·明斯基于 1951 年构建的。这是人工智能第一波热潮的开端。事实上，明斯基最初创造了“人工智能”一词。此前十年，人工智能的主要构建模块已经发展起来：阿兰·图灵发表了通用问题解决机器的数学公式；20 世纪 40 年代的全球战争推动了现代计算机的发展；对神经元工作原理的了解开始为生物大脑在微观层面上的工作原理提供线索；而以桑代克效果定律为脉络的动物心理学研究为动物智力在宏观层面上的工作原理提供了一般原则。
THE FIRST REINFORCEMENT learning computer algorithm was built in 1951 by a doctoral student at Princeton named Marvin Minsky. This was the beginning of the first wave of excitement around artificial intelligence. In fact, it was Minsky who originally coined the term artificial intelligence. The prior decade had seen the development of the main building blocks for AI: Alan Turing had published his mathematical formulation of general purpose problem-solving machines; the global war effort in the 1940s led to the development of modern computers; an understanding of how neurons worked was beginning to provide clues to how biological brains worked on the micro level; and the study of animal psychology in the vein of Thorndike’s law of effect had provided general principles for how animal intelligence worked on the macro level.
于是，马文·明斯基开始着手构建一种像桑代克动物一样学习的算法。他将自己的算法命名为随机神经模拟强化计算器，简称 SNARC。他创建了一个具有 40 个连接的人工神经网络，并训练它穿越各种迷宫。训练过程很简单：每当他的系统成功走出迷宫时，他都会强化最近激活的突触。就像桑代克用食物强化训练猫逃离迷宫一样，明斯基也在用数字强化训练人工智能逃离迷宫。
And so Marvin Minsky set out to build an algorithm that would learn like a Thorndikian animal. He named his algorithm the Stochastic Neural-Analog Reinforcement Calculator, or SNARC. He created an artificial neural network with forty connections and trained it to navigate through various mazes. The training process was simple: whenever his system successfully got out of the maze, he strengthened the recently activated synapses. Like Thorndike training a cat to escape a puzzle box with food reinforcements, Minsky was training an AI to escape mazes with numerical reinforcements.
Minsky 的 SNARC 效果不佳。该算法在随着时间的推移，它能够走出简单的迷宫，但只要遇到稍微复杂一点的情况，它就会失败。明斯基是第一个意识到，按照桑代克认为的动物学习的方式训练算法——直接强化积极结果，惩罚消极结果——是不去上班。
Minsky’s SNARC did not work well. The algorithm got better at navigating out of simple mazes over time, but whenever it faced even slightly more complex situations, it failed. Minsky was one of the first to realize that training algorithms the way that Thorndike believed animals learned—by directly reinforcing positive outcomes and punishing negative outcomes—was not going to work.
原因如下。假设我们用 Thorndike 的反复试验学习法教人工智能下跳棋。这个人工智能会先随机下棋，我们每次赢了都会给它奖励，输了就会给它惩罚。想必，如果它下足够多的跳棋，它应该会变得更好。但问题是：跳棋游戏中的强化和惩罚——输赢的结果——只发生在游戏结束时。一场游戏可能包含数百步。如果你赢了，哪些走法应该被称赞为好？如果你输了，哪些走法应该被称赞为坏？
Here’s why. Suppose we teach an AI to play checkers using Thorndike’s version of trial-and-error learning. This AI would start by making random moves, and we would give it a reward whenever it won and a punishment whenever it lost. Presumably, if it played enough games of checkers, it should get better. But here’s the problem: The reinforcements and punishments in a game of checkers—the outcome of winning or losing—occur only at the end of the game. A game can consist of hundreds of moves. If you win, which moves should get credit for being good? If you lose, which moves should get credit for being bad?
当然，这只是我们在第 4 章中看到的信用分配问题的另一个版本。当光和声音都与食物同时出现时，哪种刺激应该与食物相关联？我们已经回顾了简单的两侧对称动物用来确定这一点的技巧：掩盖（选择最强的刺激）、潜在抑制（选择新刺激）和阻断（选择之前关联过的刺激）。虽然这些解决方案在分配时间重叠的刺激之间的信用时很有用，但在分配时间分离的刺激之间的信用时却毫无用处。明斯基意识到，如果没有合理的跨时间分配信用策略，强化学习就无法发挥作用；这称为时间信用分配问题。
This is, of course, just another version of the credit assignment problem we saw in chapter 4. When a light and a sound both occur alongside food, which stimulus should get associated with food? We already reviewed tricks that simple bilaterians use to decide this: overshadowing (choosing the strongest stimulus), latent inhibition (choosing the novel stimulus), and blocking (choosing what has been associated before). While these solutions are useful when assigning credit between stimuli that overlap in time, they are useless when assigning credit between stimuli that are separated in time. Minsky realized that reinforcement learning would not work without a reasonable strategy for assigning credit across time; this is called the temporal credit assignment problem.
一种解决方案是强化或惩罚在获胜或失败之前最近发生的行为。行为和奖励之间的时间窗口越大，强化的程度就越低。这就是明斯基的 SNARC 的工作原理。但这只在时间窗口较短的情况下有效。即使在跳棋游戏中，这也是一个站不住脚的解决方案。如果下跳棋的人工智能以这种方式分配功劳，那么游戏结束时的举动总是会得到大部分功劳，而游戏开始时的举动只会得到很少的功劳。这将是愚蠢的——整个游戏可能在游戏真正获胜或失败之前，就因为一开始的一个聪明举动而获胜了。
One solution is to reinforce or punish actions that occurred recently before winning or losing. The greater the time window between an action and a reward, the less it gets reinforced. This was how Minsky’s SNARC worked. But this works only in situations with short time windows. Even in the game of checkers this is an untenable solution. If a checkers-playing AI assigned credit in this way, then the moves toward the end of the game would always get most of the credit and those toward the beginning very little. This would be dumb—the entire game might have been won on a single clever move in the beginning, long before the game was actually won or lost.
另一种解决方案是在获胜游戏结束时强化所有先前的举动（或者相反，在失败游戏结束时惩罚所有先前的举动）。你的开局失误、中间扭转局势的举动以及不可避免的结局都将得到同等的强化或惩罚，具体取决于你是赢还是输。论点是这样的：如果人工智能玩的游戏足够多，它最终将能够分辨出哪些举动是好的，哪些是坏的。
An alternative solution is to reinforce all the prior moves at the end of a winning game (or conversely, punish all the prior moves at the end of a losing game). Your opening blunder, the tide-turning move in the middle, and the inevitable finish will all get reinforced or punished equally depending on whether you won or lost. The argument goes like this: If the AI plays enough games, it will eventually be able to tell the difference between the specific moves that were good and those that were bad.
但这种解决方案也行不通。游戏配置太多，无法在合理的时间内学会哪些动作是好的。跳棋有超过五百亿亿种可能的玩法。国际象棋有超过 10 120种可能的玩法（比宇宙中的原子数量还多）。这种方法需要人工智能玩很多游戏，以至于在它成为一个相当优秀的玩家之前，我们早就死了。
But this solution also does not work. There are too many configurations of games to learn which moves are good in any reasonable amount of time. There are over five hundred quintillion possible games of checkers. There are over 10120 possible games of chess (more than the number of atoms in the universe). Such a method would require an AI to play so many games that we would all be long dead before it became an even reasonably good player.
这让我们陷入了困境。当使用强化学习训练人工智能下跳棋、走迷宫或执行任何其他任务时，我们不能仅仅强化最近的动作，也不能仅仅强化所有的动作。那么，人工智能如何通过强化来学习呢？
This leaves us stuck. When training an AI to play checkers, navigate a maze, or do any other task using reinforcement learning we cannot merely reinforce recent moves and we cannot merely reinforce all the moves. How, then, can AI ever learn through reinforcement?
明斯基早在 1961 年就发现了时间信用分配问题，但几十年来一直未得到解决。这个问题非常严重，以至于强化学习算法无法解决现实世界的问题，更不用说玩简单的跳棋游戏了。
Minsky identified the temporal credit assignment problem as far back as 1961, but it was left unsolved for decades. The problem was so severe that it rendered reinforcement learning algorithms impotent to solve real-world problems, let alone play a simple game of checkers.
然而，如今，人工强化学习算法的效果要好得多。强化学习模型在我们周围的技术中正变得越来越普遍；自动驾驶汽车、个性化广告和工厂机器人经常由它们驱动。
And yet today, artificial reinforcement learning algorithms work far better. Reinforcement learning models are becoming progressively more common in technologies all around us; self-driving cars, personalized ads, and factory robots are frequently powered by them.
我们是如何从 20 世纪 60 年代强化学习的完全无望走向如今的繁荣的？
How did we get from the complete hopelessness of reinforcement learning in the 1960s to the boom of today?
神奇的引导
A Magical Bootstrapping
1984 年，明斯基去世几十年后，一位名叫理查德·萨顿的人提交了他的最后一篇博士论文。萨顿提出了一种解决时间信用分配问题的新策略。此前六年，他在麻省大学阿默斯特分校攻读研究生，在博士后的指导下安德鲁·巴托。萨顿和巴托挖掘了强化学习的旧思想，并尝试了另一种尝试。六年的工作最终成就了萨顿的论文，他在论文中奠定了强化学习革命的思想基石之一。论文题目为：“强化学习中的时序信用分配”。
In 1984, decades after Minsky, a man named Richard Sutton submitted his final PhD dissertation. Sutton proposed a new strategy for solving the temporal credit assignment problem. He had spent the prior six years as a graduate student at UMass Amherst under the supervision of the postdoc Andrew Barto. Sutton and Barto dug up old ideas on reinforcement learning and attempted another stab at it. Six years of work culminated with Sutton’s dissertation, in which he laid one of the intellectual cornerstones for the reinforcement learning revolution. Its title: “Temporal Credit Assignment in Reinforcement Learning.”
萨顿本科学习的是心理学，而不是计算机科学，他从独特的生物学角度来解决这个问题。他并不想了解解决时间学分分配问题的最佳方法；他想了解动物解决这个问题的实际方式。萨顿的本科论文题目是“期望的统一理论”。萨顿有一种预感，期望正是之前使强化学习发挥作用的尝试中所缺少的东西。
Sutton—who had studied psychology, not computer science, as an undergraduate—tackled the problem from a uniquely biological perspective. He didn’t want to understand the best way to tackle the temporal credit assignment problem; he wanted to understand the actual way that animals solved it. Sutton’s undergraduate thesis was titled “A Unified Theory of Expectation.” And Sutton had a hunch that expectation was what was missing from previous attempts to make reinforcement learning work.
萨顿提出了一个简单但激进的想法。与其用实际奖励来强化行为，不如用预测奖励可以改变行为吗？换句话说：当人工智能系统获胜时，与其奖励它，不如当它认为自己会获胜时奖励它，结果会怎样？
Sutton proposed a simple but radical idea. Instead of reinforcing behaviors using actual rewards, what if you reinforced behaviors using predicted rewards? Put another way: Instead of rewarding an AI system when it wins, what if you reward it when the AI system thinks it is winning?
萨顿将强化学习分解为两个独立的部分：参与者和评论家。评论家在游戏的每个时刻预测获胜的可能性；它预测哪些棋盘配置很好，哪些配置不好。另一方面，参与者选择采取什么行动，并且不是在游戏结束时获得奖励，而是在评论家认为参与者的举动增加了获胜的可能性时获得奖励。参与者学习的信号不是奖励本身，而是预测奖励从一个时刻到下一个时刻的时间差异。因此，萨顿将他的方法命名为：时间差异学习。
Sutton decomposed reinforcement learning into two separate components: an actor and a critic. The critic predicts the likelihood of winning at every moment during the game; it predicts which board configurations are great and which are bad. The actor, on the other hand, chooses what action to take and gets rewarded not at the end of the game but whenever the critic thinks that the actor’s move increased the likelihood of winning. The signal on which the actor learns is not rewards, per se, but the temporal difference in the predicted reward from one moment in time to the next. Hence Sutton’s name for his method: temporal difference learning.
想象一下你在下跳棋。前九步，你和对手势均力敌。然后在第十步，你使出一些巧妙的招数扭转了局势；突然你意识到你比对手处于更有利的位置。正是在那一刻，时间差异学习信号强化了你的行动。
Imagine you are playing checkers. For the first nine moves, it is mostly neck and neck between you and your opponent. And then on the tenth move you pull off some clever maneuver that turns the tide of the game; suddenly you realize you are in a far better position than your opponent. It is that moment where a temporal difference learning signal reinforces your action.
萨顿提出，这可能解决时间信用分配问题。这将使人工智能系统能够边玩边学，而不必等到每局游戏结束。人工智能系统可以在一场漫长的跳棋游戏中强化某些动作并惩罚其他动作，无论是否赢得了或输掉了整场比赛。事实上，有时玩家在一场最终输掉的比赛中做出许多好举动，有时玩家在一场最终获胜的比赛中做出许多坏举动。
This, Sutton proposed, might solve the temporal credit assignment problem. This would enable an AI system to learn as it goes instead of having to wait until the end of each game. An AI system can reinforce some moves and punish others throughout a long game of checkers, whether or not it won or lost the overall game. Indeed, sometimes a player makes many good moves in a game he or she ultimately loses, and sometimes a player makes many bad moves in a game he or she ultimately wins.
图 6.1
Figure 6.1
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
尽管 Sutton 的方法具有直观吸引力，但我们不应期望它能奏效。Sutton 的逻辑是循环的。评论家对给定棋盘位置的获胜可能性的预测取决于参与者将采取什么未来行动（如果参与者不知道如何利用它，那么好的棋盘位置就不好）。同样，参与者决定采取什么行动取决于评论家的时间差异强化信号在强化和惩罚过去行动方面的准确性。换句话说，评论家依赖于参与者，参与者依赖于评论家。这种策略似乎从一开始就注定要失败。
Despite the intuitive appeal of Sutton’s approach, we should not expect it to work. Sutton’s logic is circular. The critic’s prediction of how likely you are to win given a board position depends on what future actions the actor will take (a good board position isn’t good if the actor doesn’t know how to take advantage of it). Similarly, the actor’s decision of what action to take depends on how accurate the critic’s temporal difference reinforcement signals have been at reinforcing and punishing past actions. In other words, the critic depends on the actor, and the actor depends on the critic. This strategy seems doomed from the start.
然而，在模拟中，Sutton 发现，通过同时训练演员和评论家，他们之间会发生神奇的引导。当然，一开始评论家经常会奖励错误的行为，而演员经常无法采取必要的行动来实现评论家的预测。但随着时间的推移，随着游戏的进行，两者会不断完善，直到它们融合在一起，产生一个能够做出惊人表现的人工智能系统智能决策。至少，萨顿的模拟结果就是这样。目前尚不清楚这在实践中是否可行。
In his simulations, however, Sutton found that by training an actor and a critic simultaneously, a magical bootstrapping occurs between them. Sure, in the beginning the critic often rewards the wrong actions, and the actor often fails to take the necessary actions to fulfill the predictions of the critic. But over time, with enough games, each refines the other until they converge to produce an AI system capable of making remarkably intelligent decisions. At least, that’s what happened in Sutton’s simulations. It wasn’t clear whether this would work in practice.
图 6.2
Figure 6.2
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
在 Sutton 致力于 TD 学习的同时，一位名叫 Gerald Tesauro 的年轻物理学家正在研究如何让 AI 系统玩西洋双陆棋。Tesauro 当时在 IBM Research，该小组后来开发了 Deep Blue（击败 Garry人类计算机系统在国际象棋中的地位与人类相当，例如国际象棋中的卡斯帕罗夫 (Kasparov) 和在《危险边缘》中击败肯·詹宁斯 (Ken Jennings) 的沃森 (Watson )。但在深蓝 (Deep Blue) 或沃森 (Watson) 之前，还有 Neurogammon。Neurogammon 是一个玩西洋双陆棋的人工智能系统，它接受了数百场专业西洋双陆棋比赛记录的训练。它不是通过反复试验来学习，而是通过尝试复制它认为人类专家会做的事情来学习。到 1989 年，Neurogammon 可以击败所有其他玩西洋双陆棋的计算机程序，但与人类相比，它显得逊色不少，甚至无法击败中级水平的玩家。
At the same time that Sutton was working on TD learning, a young physicist by the name of Gerald Tesauro was working on getting AI systems to play backgammon. Tesauro was at IBM Research, the same group that would later build Deep Blue (the program that famously beat Garry Kasparov in chess) and Watson (the program that famously beat Ken Jennings in Jeopardy!). But before Deep Blue or Watson, there was Neurogammon. Neurogammon was a backgammon-playing AI system that was trained on transcripts of hundreds of expertly played backgammon games. It learned not through trial and error but by attempting to replicate what it believed a human expert would do. By 1989, Neurogammon could beat every other backgammon-playing computer program, but it was lackluster compared to a human, unable to beat even an intermediate-level player.
当 Tesauro 偶然发现 Sutton 的 TD 学习成果时，他已经花了数年时间尝试了所有可以想象到的技术，希望让计算机能够像人类一样玩西洋双陆棋。他最辉煌的成就是 Neurogammon，这款游戏很聪明，但水平却停留在中等水平。因此 Tesauro 乐于接受新想法，甚至包括 Sutton 的激进想法，即让系统通过自己的预测进行自我教学。
By the time Tesauro stumbled on Sutton’s work on TD learning, he had spent years trying every conceivable technique to get his computer to play backgammon as well as a human. His crowning achievement was Neurogammon, which was clever but stuck at an intermediate level. And so Tesauro was open to new ideas, even Sutton’s radical idea of allowing a system to teach itself from its own predictions.
特萨罗是第一个将萨顿的想法付诸实践的人。20 世纪 90 年代初，他开始研究 TD-Gammon，这是一种使用时间差分学习来学习玩西洋双陆棋的系统。
It was Tesauro who first put Sutton’s idea to a practical test. In the early 1990s he began working on TD-Gammon, a system that learned to play backgammon using temporal difference learning.
Tesauro 对此持怀疑态度。Neurogammon 曾接受过人类专家玩家的教学——向它展示了最佳动作——而 TD-Gammon 则完全通过反复试验来学习，这需要它自己发现最佳动作。然而，到 1994 年，用 Tesauro 自己的话来说，TD-Gammon 取得了“真正的令人震惊的表现水平。”它不仅击败了 Neurogammon，而且与世界上一些最优秀的西洋双陆棋选手一样优秀。虽然 Sutton 证明了时间差分学习在理论上是可行的，但 Tesauro 证明了它在实践中是可行的。在随后的几十年里，TD 学习将用于训练 AI 系统以人类水平的技能完成许多任务，从玩 Atari 游戏到在自动驾驶汽车中变换车道。
Tesauro was skeptical. Neurogammon had been taught with examples from expert human players—it was shown the best moves—while TD-Gammon learned solely from trial and error, requiring it to discover the best moves on its own. And yet, by 1994, TD-Gammon achieved, in Tesauro’s own words, a “truly staggering level of performance.” It not only blew Neurogammon away but was as good as some of the best backgammon players in the world. While it was Sutton who proved that temporal difference learning worked in theory, it was Tesauro who proved it worked in practice. In the decades that followed, TD learning would be used to train AI systems to do many tasks with human-level skill, from playing Atari games to changing lanes in self-driving cars.
然而，真正的问题是，TD 学习是否只是一种碰巧奏效的巧妙技术，还是一种捕捉到智能本质基本特征的技术。TD 学习是一项技术发明，还是如 Sutton 所希望的那样，是一种古老的技术，在进化过程中偶然发现，并很久以前融入动物大脑，使强化学习发挥作用？
The real question, however, was whether TD learning was merely a clever technique that happened to work or a technique that captured something fundamental about the nature of intelligence. Was TD learning a technological invention, or was it, as Sutton had hoped, an ancient technique that evolution had stumbled upon and long ago weaved into animal brains to make reinforcement learning work?
多巴胺的重大改造
The Grand Repurposing of Dopamine
尽管萨顿曾希望他的想法与大脑之间存在联系，但发现这种联系的却是他的学生彼得·戴扬。在圣地亚哥的索尔克研究所，戴扬和他的博士后同事里德·蒙塔古确信大脑可以实现某种形式的 TD 学习。20 世纪 90 年代，在特萨罗的 TD-Gammon 取得成功的鼓舞下，他们开始在不断增长的神经科学数据中寻找证据。
While Sutton had hoped there was a connection between his idea and the brain, it was one of his students, Peter Dayan, who found it. At the Salk Institute in San Diego, Dayan and his fellow postdoc Read Montague were convinced that brains implemented some form of TD learning. In the 1990s, emboldened by the success of Tesauro’s TD-Gammon, they went hunting for evidence in the ever-growing mound of neuroscience data.
他们知道从哪里开始。任何试图了解强化学习如何在脊椎动物大脑中发挥作用的尝试都必须从我们已经见过的一种小神经调节剂开始：多巴胺。
They knew where to start. Any attempt to understand how reinforcement learning works in vertebrate brains surely had to begin with a little neuromodulator we have already seen: dopamine.
所有脊椎动物的中脑深处都有一小簇多巴胺神经元。这些神经元虽然数量不多，但会将输出信号发送到大脑的许多区域。20 世纪 50 年代，研究人员发现，如果你将电极插入老鼠的大脑并刺激这些多巴胺神经元，老鼠几乎可以做任何事情。如果你每隔几次老鼠推杠杆就刺激这些神经元，老鼠每小时就会推杠杆超过 5000 次连续二十四小时。事实上，如果让老鼠在释放多巴胺的杠杆和吃食物之间做出选择，它们会选择杠杆。老鼠会无视食物，让自己挨饿有利于多巴胺刺激。
Deep within the midbrain of all vertebrates is a small cluster of dopamine neurons. These neurons, while few in number, send their output to many regions of the brain. In the 1950s researchers discovered that if you put an electrode into the brain of a rat and stimulate these dopamine neurons, you can get a rat to do pretty much anything. If you stimulate these neurons every few times a rat pushes a lever, the rat will push this lever over five thousand times an hour for twenty-four hours straight. In fact, if given the choice between a dopamine-releasing lever and eating food, rats will choose the lever. Rats will ignore food and starve themselves in favor of dopamine stimulation.
这种效应也存在于鱼类中。鱼会回到它被给予多巴胺的地方，并继续这样做，即使这些区域与它通常会避免的不愉快的事情有关（比如反复从水中取出）。
This effect is also found in fish. A fish will return to places where it is given dopamine and continue doing so even if those areas are paired with unpleasant things it usually avoids (like being repeatedly removed from water).
事实上，大多数滥用药物——酒精、可卡因、尼古丁——都是通过触发多巴胺的释放起作用的。从鱼到老鼠、猴子到人类，所有脊椎动物都容易对多巴胺上瘾。比如能增强多巴胺的化学物质。
Indeed, most drugs of abuse—alcohol, cocaine, nicotine—work by triggering the release of dopamine. All vertebrates, from fish to rats to monkeys to humans, are susceptible to becoming addicted to such dopamine-enhancing chemicals.
多巴胺无疑与强化有关，但具体如何关联尚不清楚。最初的解释是多巴胺是大脑的愉悦信号；动物重复激活多巴胺神经元的行为是因为感觉良好。这在桑代克最初的试错学习概念中是有道理的，即重复导致令人满意结果的行为的过程。但我们在第 3 章中已经看到，多巴胺不会产生快感。它与喜欢无关，而与欲望有关。那么为什么多巴胺如此具有强化作用呢？
Dopamine was undeniably related to reinforcement, but how exactly was not so clear. The original interpretation was that dopamine was the brain’s pleasure signal; animals repeated behaviors that activated dopamine neurons because it felt good. This made sense in Thorndike’s original concept of trial-and-error learning as a process of repeating behaviors that led to satisfying outcomes. But we already saw in chapter 3 that dopamine does not produce pleasure. It is less about liking and more about wanting. So then why was dopamine so reinforcing?
要知道多巴胺发出的信号，唯一的办法就是测量信号。直到 20 世纪 80 年代，技术才足够先进，科学家才能够做到这一点。德国神经科学家沃尔夫拉姆·舒尔茨是第一个测量单个多巴胺神经元活动的人。
The only way to know what dopamine is signaling is to, well, measure the signal. It wasn’t until the 1980s that technology was advanced enough for scientists to do this. A German neuroscientist named Wolfram Schultz was the first to measure the activity of individual dopamine neurons.
舒尔茨设计了一个简单的实验来探究多巴胺和强化之间的关系。舒尔茨向猴子展示不同的提示（例如几何图形的图片），然后几秒钟后将一些糖水送入它们的嘴里。
Schultz devised a simple experiment to probe the relationship between dopamine and reinforcement. Schultz showed monkeys different cues (such as pictures of a geometric shape) and then a few seconds later delivered some sugar water into their mouths.
果然，即使在这个简单的奖励预测任务中，我们也能立即发现多巴胺并不是桑代克所说的令人满意的结果的信号——它不是愉悦或效价的信号。起初，多巴胺神经元的反应确实像效价信号一样，每当饥饿的猴子得到糖水时，它就会特别兴奋。但经过几次试验后，多巴胺神经元不再对奖励本身做出反应，而是只对预测线索做出反应。
Sure enough, even in this simple reward-prediction task, it was immediately clear that dopamine was not a signal for Thorndike’s satisfying outcomes—it was not a signal for pleasure or valence. At first, dopamine neurons did respond like a valence signal, getting uniquely excited whenever a hungry monkey got sugar water. But after a few trials, dopamine neurons stopped responding to the reward itself and instead responded only to the predictive cue.
当猴子看到一张图片时，它们的多巴胺神经元会兴奋，而猴子知道这张图片会给它们带来糖水，但几分钟后，当猴子喝糖水时，它们的多巴胺神经元并没有偏离基线活动水平。那么，也许多巴胺实际上是一种惊喜的信号？也许只有当事件偏离预期时，比如一张令人惊讶的图片突然出现或糖水突然送来，多巴胺才会兴奋？
When a picture popped up that monkeys knew would lead to sugar, their dopamine neurons got excited, but when these monkeys got sugar water a few moments later, their dopamine neurons did not deviate from their baseline level of activity. Perhaps, then, dopamine was actually a signal for surprise? Perhaps dopamine got excited only when events deviated from expectations, like a surprising picture popping up or a surprsing delivery of sugar water?
当舒尔茨进行额外的实验时，很明显这种“多巴胺作为惊喜”的想法是错误的。当他的一只猴子学会在呈现特定图片后期待糖水时，舒尔茨再次呈现了这张奖励预测图片，但没有给糖。在这种情况下，尽管惊喜程度相同，但多巴胺活性急剧下降。虽然呈现意外奖励会增加多巴胺活性，但没有预期奖励会降低多巴胺活性。*
When Schultz performed additional experiments, it became clear that this “dopamine as surprise” idea was wrong. Once one of his monkeys had learned to expect sugar water after a specific picture was presented, Schultz again presented this reward-predicting picture but didn’t give sugar. In this case, despite an equal amount of surprise, dopamine activity dramatically declined. While the presentation of an unexpected reward increases dopamine activity, the omission of an expected reward decreases dopamine activity.*
图 6.3：多巴胺神经元对预测线索、奖励和遗漏的反应
Figure 6.3: Responses of dopamine neurons to predictive cues, rewards, and omissions
图为 Max Bennett
Figure by Max Bennett
舒尔茨对这些结果感到困惑。多巴胺是一种什么信号？如果不是为了效价、愉悦或惊讶，那又是什么？为什么多巴胺活动会从奖励转向奖励的预测线索？为什么当预期奖励被忽略时，多巴胺活动会下降？
Schultz was confused by these results. What was dopamine a signal for? If not for valence or pleasure or surprise, then what? Why did dopamine activity shift from rewards to the predictive cues for rewards? Why did dopamine activity decline when expected rewards were omitted?
多年来，神经科学界一直不确定如何解释舒尔茨的数据，这种奇怪的现象在一种古老的神经元的点击和停顿中暴露无遗。
For many years, the neuroscience community was unsure how to interpret Schultz’s data, an oddity laid bare in the clicks and pauses of an ancient type of neuron.
直到十年后，这个问题才得到解决。事实上，十年后，Dayan 和 Montague 开始在文献中寻找大脑实施某种形式的 TD 学习的线索。当他们最终遇到 Schultz 的数据时，他们立即知道他们看到了什么。Schultz 在猴子身上发现的多巴胺反应与 Sutton 的完全一致时间差异学习信号。舒尔茨猴子的多巴胺神经元因预测线索而兴奋，因为这些线索导致预测未来奖励的增加（正时间差异）；多巴胺神经元不受预期奖励的影响因为预测的未来奖励没有变化（没有时间差异）；而当预期奖励被省略时，多巴胺神经元活动会减少，因为预测的未来奖励会减少（负时间差异）。
It wasn’t until a decade later that it was solved. Indeed, it was a decade later when Dayan and Montague began scouring the literature for clues that brains implement some form of TD learning. When they eventually came across Schultz’s data, they immediately knew what they were seeing. The dopamine responses that Schultz found in monkeys aligned exactly with Sutton’s temporal difference learning signal. Dopamine neurons in Schultz’s monkeys got excited by predictive cues because these cues led to an increase in predicted future rewards (a positive temporal difference); dopamine neurons were unaffected by the delivery of an expected reward because there was no change in predicted future reward (no temporal difference); and dopamine-neuron activity decreased when expected rewards were omitted because there was a decrease in predicted future rewards (a negative temporal difference).
多巴胺反应的细微差别也与时间差异信号完全一致。例如，舒尔茨发现，预测四秒内有食物的提示比预测四秒内有食物的提示触发了更多的多巴胺十六秒内食物。这被称为折扣，Sutton 也将其融入到他的 TD 学习信号中；折扣促使 AI 系统（或动物）选择尽早而不是推迟获得奖励的行为。
Even the subtleties of dopamine responses aligned exactly with a temporal difference signal. For example, Schultz found that a cue that predicts food in four seconds triggers more dopamine than a cue that predicts food in sixteen seconds. This is called discounting, something Sutton also incorporated into his TD-learning signal; discounting drives AI systems (or animals) to choose actions that lead to rewards sooner rather than later.
甚至多巴胺对与 TD 学习信号一致的概率的反应方式——以 75% 的概率预测食物的提示比以 25% 的概率预测食物的提示触发更多的多巴胺。
Even the way dopamine responds to probabilities aligned with a TD-learning signal—a cue that predicts food with a 75 percent probability triggers more dopamine than a cue that predicts food with a 25 percent probability.
多巴胺不是奖励的信号，而是强化的信号。正如 Sutton 所发现的，强化和奖励必须分离，才能使强化学习发挥作用。为了解决时间信用分配问题，大脑必须根据预测的未来奖励的变化来强化行为，而不是实际奖励。这就是为什么动物会对多巴胺释放行为上瘾，尽管这种行为并不令人愉快，这就是为什么多巴胺反应会迅速将其激活转移到动物预测即将到来的奖励的时刻，而不是奖励本身。
Dopamine is not a signal for reward but for reinforcement. As Sutton found, reinforcement and reward must be decoupled for reinforcement learning to work. To solve the temporal credit assignment problem, brains must reinforce behaviors based on changes in predicted future rewards, not actual rewards. This is why animals get addicted to dopamine-releasing behaviors despite it not being pleasurable, and this is why dopamine responses quickly shift their activations to the moments when animals predict upcoming reward and away from rewards themselves.
1997 年，Dayan 和 Montague 与 Schultz 合作发表了一篇具有里程碑意义的论文，题为“预测和奖励的神经基础”。至今，这一发现仍是人工智能和神经科学之间最著名、最美丽的合作之一。受 Sutton 对大脑运作方式的思考启发，一种策略成功克服了人工智能中的实际挑战，这反过来又帮助我们解读了有关大脑的神秘数据。神经科学为人工智能提供信息，人工智能为神经科学提供信息。
In 1997 Dayan and Montague published a landmark paper, coauthored with Schultz, titled “A Neural Substrate of Prediction and Reward.” To this day, this discovery represents one of the most famous and beautiful partnerships between AI and neuroscience. A strategy inspired by how Sutton thought the brain might work turned out to successfully overcome practical challenges in AI, and this in turn helped us interpret mysterious data about the brain. Neuroscience informing AI, and AI informing neuroscience.
大多数记录多巴胺神经元活动的研究都是在哺乳动物身上进行的，但有充分的理由相信多巴胺的这些特性也适用于鱼类。鱼类和哺乳动物大脑中的多巴胺系统回路大致相同，在鱼类、大鼠和猴子，和人类大脑。相比之下，在线虫或其他简单的两侧对称动物的多巴胺神经元中没有发现 TD 学习信号。*
Most studies that record the activity of dopamine neurons have been done in mammals, but there is every reason to believe these properties of dopamine extend to fish as well. The dopamine-system circuitry is largely the same in both fish and mammal brains, and the same TD-learning signals have been found in the brain structures of fish, rat, monkey, and human brains. In contrast, no TD-learning signals have been found in the dopamine neurons of nematodes or other simple bilaterians.*
在早期的两侧对称动物中，多巴胺是附近有好东西的信号——一种原始的渴望。*然而，在进化为脊椎动物的过程中，这种附近有好东西的信号被精心设计，不仅可以触发渴望状态，还可以传达精确计算的时间差学习信号。事实上，多巴胺是一种神经调节剂，进化将其重塑为时间差学习信号，这是有道理的，因为附近奖励的信号是最接近预测未来奖励的指标。因此，多巴胺从附近有好东西的信号转变为在 10 秒内有 35% 的机会发生一些很棒的事情的信号。从最近检测到的食物的模糊平均值转变为不断波动、精确测量和精心计算的预测未来奖励信号。
In early bilaterians, dopamine was a signal for good things nearby—a primitive version of wanting.* In the transition to vertebrates, however, this good-things-are-nearby signal was elaborated to not only trigger a state of wanting but also to communicate a precisely computed temporal difference learning signal. Indeed, it makes sense that dopamine was the neuromodulator that evolution reshaped into a temporal difference learning signal, as the signal for nearby rewards it was the closest thing to a measure of predicted future reward. And so, dopamine was transformed from a good-things-are-nearby signal to a there-is-a-35 percent-chance-of-something-awesome-happening-in-exactly-ten-seconds signal. Repurposed from a fuzzy average of recently detected food to an ever fluctuating, precisely measured, and meticulously computed predicted-future-reward signal.
解脱、失望和时机的出现
The Emergence of Relief, Disappointment, and Timing
从 TD 学习的古老种子中萌发了几种智力特征。其中两种——失望和解脱——是如此熟悉，以至于几乎从人们的视野中消失，它们如此普遍，以至于人们很容易忽略它们并不总是存在的不可避免的事实。失望和解脱都是大脑通过预测未来奖励来学习的突发特性。事实上，如果不能准确预测未来的奖励，当奖励没有发生时就不会有失望。如果不能准确预测未来的痛苦，当痛苦没有发生时就不会有解脱。
From the ancient seed of TD learning sprouted several features of intelligence. Two of these—disappointment and relief—are so familiar that they almost disappear from view, so ubiquitous that it is easy to miss the unavoidable fact that they did not always exist. Both disappointment and relief are emergent properties of a brain designed to learn by predicting future rewards. Indeed, without an accurate prediction of a future reward, there can be no disappointment when it does not occur. And without an accurate prediction of future pain, there can be no relief when it does not occur.
考虑一下以下鱼通过反复试验进行学习的任务。如果你打开一盏灯，然后在五秒钟后，如果鱼没有游到鱼缸的另一边，就轻轻地电击它，它会学会在你打开灯时自动游到鱼缸的另一边。这似乎是简单的反复试验学习，对吧？不幸的是不是。脊椎动物执行这种任务（称为回避任务）的能力长期以来一直是动物心理学家争论的焦点。
Consider the following task of a fish learning through trial and error. If you turn on a light and then after five seconds gently zap the fish if it does not swim to the opposite side of a tank, it will learn to automatically swim to the opposite side of the tank whenever you turn the light on. Seems like straightforward trial-and-error learning right? Unfortunately not. The ability of vertebrates to perform this type of task—called an avoidance task—has long been a source of debate among animal psychologists.
桑代克会如何解释鱼的这种能力呢？当桑代克的一只猫终于从谜题盒中出来时，食物奖励的存在强化了猫的行为。但当我们的鱼游到安全的地方时，没有预期的电击却强化了鱼的行为。没有某种东西怎么会起到强化作用呢？
How would Thorndike have explained a fish’s ability to do this? When one of Thorndike’s cats finally got out of a puzzle box, it was the presence of food rewards that reinforced the cat’s actions. But when our fish swam to the safe location, it was the omission of a predicted shock that reinforced the fish’s actions. How can the absence of something be reinforcing?
答案是，不给予预期惩罚本身就是一种强化，是一种缓解。而不给予预期奖励本身就是一种惩罚，是一种失望。这就是为什么当不给予食物时，舒尔茨的多巴胺神经元活动会减少。他观察的是失望的生物学表现——大脑的惩罚信号对未来奖励的预测失败。
The answer is that the omission of an expected punishment is itself reinforcing; it is relieving. And the omission of an expected reward is itself punishing; it is disappointing. This is why the activity of Schultz’s dopamine neurons decreased when food was omitted. He was observing the biological manifestation of disappointment—the brain’s punishment signal for a failed prediction of future reward.
事实上，你可以训练脊椎动物，甚至是鱼，不仅通过奖励和惩罚来执行任意动作，还可以通过省略预期的奖励或惩罚。对一些人来说，一块意外的甜点（奖励）与学校意外放假（省去一些预期但不喜欢的东西）一样具有强化作用。
Indeed, you can train vertebrates, even fish, to perform arbitrary actions not only with rewards and punishments but also with the omission of expected rewards or punishments. To some, a surprise piece of dessert (a reward) is as reinforcing as a surprise day off from school (the omission of something expected but disliked).
另一方面，线虫无法学会任意通过省略奖励来学习行为。即使是独立进化出许多智力的螃蟹和蜜蜂，也无法通过省略事物来学习。*
A nematode, on the other hand, cannot learn to perform arbitrary behaviors through the omission of rewards. Even crabs and honeybees, who independently evolved many intellectual faculties, are not be able to learn from omission of things.*
在脊椎动物和无脊椎动物之间的智力鸿沟中，我们发现了另一种常见的智力特征，这种特征也出现在时间差学习及其对应的失望和解脱中。如果我们仔细观察我们的鱼学习游到特定位置以躲避电击，我们会发现一些了不起的事情。当灯亮时，鱼不会立即冲向安全的地方。相反，它会悠闲地忽略灯光，直到五秒钟间隔结束前，然后迅速冲向安全的地方。在这个简单的任务中，鱼不仅学会了做什么，还学会了什么时候做；鱼知道电击发生在正确的时间。点亮后五秒。
In this intellectual divide between vertebrates and invertebrates we find another familiar feature of intelligence, one that also emerges from TD learning and its counterparts of disappointment and relief. If we looked closely at our fish learning to swim to specific locations to avoid a zap, we would observe something remarkable. When the light turns on, the fish does not immediately dash to safety. Instead, it leisurely ignores the light until just before the end of the five-second interval and then rapidly dashes to safety. In this simple task, fish learn not only what to do but when to do it; fish know that the shock occurs precisely five seconds after light.
许多生命形式都有追踪时间流逝的机制。细菌、动物和植物都有昼夜节律钟来追踪一天的周期。但脊椎动物在测量时间的精确度方面是独一无二的。脊椎动物可以记住一个事件恰好发生在另一个事件发生五秒之后。相比之下，像蛞蝓和扁虫这样的简单两侧对称动物完全无法学习精确的时间事件之间的间隔。事实上，像蛞蝓这样的简单两侧对称动物甚至无法学会将相隔两秒以上的事件联系起来，更不用说知道一件事恰好发生在另一件事之后五秒。即使是螃蟹和蜜蜂等高级无脊椎动物也无法学习事件之间的精确时间间隔。
Many forms of life have mechanisms to track the passage of time. Bacteria, animals, and plants all have circadian clocks to track the cycle of the day. But vertebrates are unique in the precision with which they can measure time. A verterbate can remember that one event occurs precisely five seconds after another event. In contrast, simple bilaterians like slugs and flatworms are entirely unable to learn the precise time intervals between events. Indeed, simple bilaterians like slugs cannot even learn to associate events separated by more than two seconds, let alone learn that one thing happens exactly five seconds after another. Even the advanced invertebrates like crabs and bees are unable to learn precise time intervals between events.
时间差分学习、失望、释然和时间感知都是相互关联的。精确的时间感知是从遗漏中学习、知道何时触发失望或释然，从而使时间差分学习发挥作用的必要因素。没有时间感知，大脑就无法知道某件事是被遗漏了还是根本就没有发生；我们的鱼会知道光与电击有关，但不知道电击何时发生。在电击的风险过去很久之后，我们的鱼会在光的面前畏缩不前，对自己的安全视而不见。只有通过内在的时钟，鱼才能预测电击发生的确切时刻，因此，如果遗漏了，它应该在什么时候获得缓解的多巴胺爆发。
TD learning, disappointment, relief, and the perception of time are all related. The precise perception of time is a necessary ingredient to learn from omission, to know when to trigger disappointment or relief, and thereby to make TD learning work. Without time perception, a brain cannot know whether something was omitted or simply hasn’t happened yet; our fish would know that the light was associated with a zap but not when it should occur. Our fish would cower in fear in the presence of the light long after the risk of the zap had passed, blind to its own safety. It is only with an inner clock that fish can predict the exact moment the zap would occur and thus, if omitted, the exact moment it deserves a relieving dopamine burst.
基底神经节
The Basal Ganglia
我最喜欢的大脑部分是叫做基底神经节的结构。
My favorite part of the brain is a structure called the basal ganglia.
对于大多数大脑结构，人们了解得越多，就越不理解它们——简化的框架在混乱的复杂性（生物系统的标志）的重压下崩溃。但基底神经节却不同。它的内部线路展现出令人着迷的美丽设计，展现出有序的计算和功能。就像人们可能对进化能够构造出如此对称和优雅的眼睛感到敬畏一样，人们同样可能对进化能够构造出同样具有对称性和优雅性的基底神经节感到敬畏。
For most brain structures, the more one learns about them, the less one understands them—simplified frameworks crumble under the weight of messy complexity, the hallmark of biological systems. But the basal ganglia is different. Its inner wiring reveals a mesmerizing and beautiful design, exposing an orderly computation and function. As one might feel awe that evolution could construct an eye, with such symmetry and elegance, one could equally feel awe that evolution could construct the basal ganglia, also endowed with its own symmetry and elegance.
基底神经节位于皮层和丘脑之间（见本书前几页的图）。基底神经节的输入来自皮层、丘脑和中脑，使基底神经节能够监测动物的行为和外部环境。然后，信息流经基底神经节内错综复杂的子结构，分支和合并、转换和排列，直到到达基底神经节的输出核，该核包含数千到数百万个抑制神经元，这些神经元向脑干的运动中枢发送大量强大的连接。基底神经节的这个输出核默认处于激活状态。脑干的运动回路不断受到基底神经节的抑制和控制。只有当基底神经节中的特定神经元关闭时，脑干中的特定运动回路才会不受激活控制。基底神经节因此处于控制和解除特定动作的永久状态，充当动物行为的整体操纵者。
The basal ganglia is wedged between the cortex and the thalamus (see the figure in the first pages of this book). The input to the basal ganglia comes from the cortex, thalamus, and midbrain, enabling the basal ganglia to monitor an animal’s actions and external environment. Information then flows through a labyrinth of substructures within the basal ganglia, branching and merging, transforming and permuting until it reaches the basal ganglia’s output nucleus, which contains thousands to millions of inhibitory neurons that send massive and powerful connections to motor centers in the brainstem. This output nucleus of the basal ganglia is, by default, activated. The motor circuits of the brainstem are constantly being suppressed and gated by the basal ganglia. It is only when specific neurons in the basal ganglia turn off that specific motor circuits in the brainstem are ungated from activation. The basal ganglia is thereby in a perpetual state of gating and ungating specific actions, operating as a global puppeteer of an animal’s behavior.
基底神经节的功能对我们的生活至关重要。帕金森病的典型症状是无法发起运动。患者会在椅子上坐好几分钟，然后才能鼓起勇气坐起来。帕金森病的这种症状主要是由于基底神经节的破坏而出现的，它处于一种永久控制所有动作的状态，从而使患者无法发起哪怕是最简单的动作。
The functioning of the basal ganglia is essential to our lives. The cononical symptom of Parkinson’s disease is the inability to initiate movement. Patients will sit in a chair for many minutes before they can muster the will to even sit up. This symptom of Parkinson’s disease primarily emerges due to disruption of the basal ganglia, leaving it in a perpetual state of gating all actions, thereby depriving patients of the ability to initiate even the simplest of movements.
基底神经节执行哪些计算？它如何使用有关动物行为和外部环境的传入信息来决定哪些行为需要控制（防止发生）以及哪些行为需要放开（允许发生）？
What is the computation performed by the basal ganglia? How does it use incoming information about an animal’s actions and external environment to decide which actions to gate (prevent from occuring) and which actions to ungate (allow to occur)?
除了接收有关动物行为和外部环境的信息外，基底神经节还接收来自多巴胺神经元簇的输入。每当这些多巴胺神经元兴奋时，基底神经节就会迅速充满多巴胺；每当这些多巴胺神经元受到抑制时，基底神经节就会迅速缺乏多巴胺。基底神经节内的突触具有不同的多巴胺受体，每个受体都以独特的方式做出反应；这些波动的多巴胺水平会增强或削弱特定的突触，从而改变基底神经节处理输入的方式。
In addition to receiving information about an animal’s actions and external environment, the basal ganglia also receives input from a cluster of dopamine neurons. Whenever these dopamine neurons get excited, the basal ganglia is rapidly flooded with dopamine; whenever these dopamine neurons are inhibited, the basal ganglia is rapidly starved of dopamine. The synapses within the basal ganglia have different dopamine receptors, each responding in unique ways; these fluctuating levels of dopamine strengthen and weaken specific synapses, modifying how the basal ganglia processes input.
随着神经科学家追踪基底神经节的回路，它的功能变得相当清晰。基底神经节学会重复动作最大化多巴胺释放。通过基底神经节，导致多巴胺释放的行为更有可能发生（基底神经节会取消这些行为），而导致多巴胺抑制的行为则不太可能发生（基底神经节会控制这些行为）。听起来熟悉吗？基底神经节在某种程度上是 Sutton 的“演员”——一种旨在重复导致强化的行为并抑制导致惩罚的行为的系统。
As neuroscientists traced the circuitry of the basal ganglia, its function became quite clear. The basal ganglia learns to repeat actions that maximize dopamine release. Through the basal ganglia, actions that lead to dopamine release become more likely to occur (the basal ganglia ungates those actions), and actions that lead to dopmaine inhibition become less likely to occur (the basal ganglia gates those actions). Sound familiar? The basal ganglia is, in part, Sutton’s “actor”—a system designed to repeat behaviors that lead to reinforcement and inhibit behaviors that lead to punishment.
值得注意的是，人类大脑和七鳃鳗大脑的基底神经节回路几乎完全相同，这两个物种的共同祖先都是 5 亿多年前最早的脊椎动物。子簇、神经元类型和整体功能似乎都相同。在早期脊椎动物的大脑中出现了基底神经节，这是强化学习的生物学中心。
Remarkably, the circuitry of the basal ganglia is practically identical between a human brain and a lamprey fish brain, two species whose shared ancestors were the first vertebrates over 500 million years ago. The various subclusters, the types of neurons, and the overall function seem to be the same. In the brain of early vertebrates emerged the basal ganglia, the biological locus of reinforcement learning.
强化学习并非源自基底神经节的单独作用，而是源自基底神经节与另一个独特的脊椎动物结构——下丘脑（位于前脑底部的一个小结构）之间的古老相互作用。
Reinforcement learning emerged not from the basal ganglia acting alone, but from an ancient interplay between the basal ganglia and another uniquely vertebrate structure called the hypothalamus, which is a small structure at the base of the forebrain.
在脊椎动物的大脑中，多巴胺的释放最初由下丘脑控制。下丘脑中存在着从祖先两侧对称动物的价态传感装置中遗传下来的价态神经元。当你感到寒冷时，是你的下丘脑引发了颤抖，让你享受温暖；就像当你感到炎热时，是你的下丘脑引发了出汗，让你享受寒冷一样。当你的身体需要卡路里时，是你的下丘脑检测到血液中的饥饿信号，让你感到饥饿。早期两侧对称动物的正价态食物敏感神经元的功能与下丘脑中的正价态食物敏感神经元一样，当你饥饿时对食物反应强烈，而当你饱腹时对食物的反应较弱。这就是为什么你前一刻还在为披萨垂涎三尺，但饱餐一顿后，十分钟后就再也不想吃披萨了。
In vertebrate brains, dopamine release is initially controlled by the hypothalamus. It is the hypothalamus that houses valence neurons inherited from the valence sensory apparatus of ancestral bilaterians. When you are cold, it is your hypothalamus that triggers shivering and that makes you enjoy warmth; just as when you are hot, it is your hypothalamus that triggers sweating and that makes you enjoy the cold. When your body needs calories, it is your hypothalamus that detects hunger signals in your bloodstream and that makes you hungry. The positive valence food-sensitive neurons in early bilaterians functioned just as the positive valence food-sensitive neurons in your hypothalamus do, becoming highly responsive to food when you are hungry and less responsive to food when you are full. This is why you will be salivating over pizza one moment, but then after engorging yourself want absolutely nothing to do with pizza just ten minutes later.
换句话说，下丘脑原则上只是早期两侧对称动物控制大脑的更复杂版本；它将外部刺激分为好与坏，并触发对每种刺激的反射反应。下丘脑的价神经元连接到同一簇多巴胺神经元，该簇将多巴胺传播到整个基底神经节。当下丘脑高兴时，它会向基底神经节注入多巴胺，而当它心情不好时，它会剥夺基底神经节的多巴胺。基底神经节的多巴胺。因此，从某种意义上来说，基底神经节是一个学生，总是试图满足其模糊但严厉的下丘脑评判者。
In other words, the hypothalamus is, in principle, just a more sophisticated version of the steering brain of early bilaterians; it reduces external stimuli to good and bad and triggers reflexive responses to each. The valence neurons of the hypothalamus connect to the same cluster of dopamine neurons that propogates dopamine throughout the basal ganglia. When the hypothalamus is happy, it floods the basal ganglia with dopamine, and when it is upset, it deprives the basal ganglia of dopamine. And so, in some ways, the basal ganglia is a student, always trying to satisfy its vague but stern hypothalamic judge.
下丘脑不会因预测线索而兴奋；只有当它真正得到它想要的东西时才会兴奋——饥饿时得到食物，寒冷时得到温暖。下丘脑是实际奖励的决定者；在我们的人工智能玩西洋双陆棋的比喻中，下丘脑告诉大脑它是赢了还是输了游戏，但不告诉大脑它在游戏展开时表现如何。
The hypothalamus doesn’t get excited by predictive cues; it gets excited only when it actually gets what it wants—food when hungry, warmth when cold. The hypothalamus is the decider of actual rewards; in our AI-playing-backgammon metaphor, the hypothalamus tells the brain whether it won or lost the game but not how well it is doing as the game is unfolding.
图 6.4：第一个脊椎动物大脑设计的简化框架
Figure 6.4: A simplified framework for the design of the first vertebrate brain
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
但正如明斯基在 20 世纪 50 年代尝试制作强化学习算法时发现的那样，如果大脑只从实际奖励中学习，它们就永远无法做出任何如此智能的事情。它们会受到时间信用分配问题的困扰。那么多巴胺是如何从实际奖励的价态信号转变为预测未来奖励变化的时间差异信号的呢？
But as Minsky found with his attempts to make reinforcement learning algorithms in the 1950s, if brains learned only from actual rewards, they would never be able to do anything all that intelligent. They would suffer from the problem of temporal credit assignment. So then how is dopamine transformed from a valence signal for actual rewards to a temporal difference signal for changes in predicted future reward?
在所有脊椎动物中，基底神经节内都存在着神秘的平行回路马赛克，其中一个回路向下流向运动回路和门控运动，另一个回路则流回多巴胺神经元直接参与基底神经节的功能。基底神经节功能的一个主要理论是，这些平行回路实际上是 Sutton 的演员-评论家系统，用于实现时间差异学习。一个回路是“演员”，学习重复触发多巴胺释放的行为；另一个回路是“评论家”，学习预测未来的奖励并触发自己的多巴胺激活。
In all vertebrates, there is a mysterious mosaic of parallel circuits within the basal ganglia, one that flows down to motor circuits and gates movement, and another that flows back toward dopamine neurons directly. One leading theory of basal ganglia function is that these parallel circuits are literally Sutton’s actor-critic system for implementing temporal difference learning. One circuit is the “actor,” learning to repeat the behaviors that trigger dopamine release; the other circuit is the “critic,” learning to predict future rewards and trigger its own dopamine activation.
在我们的比喻中，基底神经节学生最初仅从下丘脑法官那里学习，但随着时间的推移，它学会了自我判断，在下丘脑给出任何反馈之前就知道自己何时犯了错误。这就是为什么多巴胺神经元最初在获得奖励时做出反应，但随着时间的推移，它们的激活会转向预测线索。这也是为什么收到你知道会收到的奖励不会触发多巴胺释放；基底神经节的预测抵消了下丘脑的兴奋。
In our metaphor, the basal ganglian student initially learns solely from the hypothalamic judge, but over time learns to judge itself, knowing when it makes a mistake before the hypothalamus gives any feedback. This is why dopamine neurons initially respond when rewards are delivered, but over time shift their activation toward predictive cues. This is also why receiving a reward that you knew you were going to receive doesn’t trigger dopamine release; predictions from the basal ganglia cancel out the excitement from the hypothalamus.
基底神经节的回路保存完好，最早出现在早期脊椎动物的微小大脑中，并维持了五亿年，这似乎是萨顿的演员-评论家系统的生物表现。萨顿发现了进化在五亿多年前就已发现的窍门。
The beautifully conserved circuitry of the basal ganglia, first emerging in the minuscule brain of early vertebrates and maintained for five hundred million years, seems to be the biological manifestation of Sutton’s actor-critic system. Sutton discovered a trick that evolution had already stumbled upon over five hundred million years ago.
TD 学习、脊椎动物基底神经节的布线、多巴胺反应的特性、学习精确时间间隔的能力以及从遗漏中学习的能力都交织在使反复试验学习发挥作用的相同机制中。
TD learning, the wiring of vertebrate basal ganglia, the properties of dopamine responses, the ability to learn precise time intervals, and the ability to learn from omissions are all interwoven into the same mechanisms for making trial-and-error learning work.
7
7
模式识别问题
The Problems of Pattern Recognition
五亿年前，现存所有脊椎动物的鱼类祖先——每只鸽子、鲨鱼、老鼠、狗，当然还有人类的祖母，身长只有一英寸——在不知不觉中游向危险。她游过寒武纪半透明的水下植物，轻轻地穿梭在厚厚的海藻状茎之间。她在寻找珊瑚幼虫，这些幼虫是生活在海洋中的无脑动物的富含蛋白质的后代。她不知道，她也在被猎杀。
FIVE HUNDRED MILLION years ago, the fish-like ancestor of every vertebrate alive today—the inch-long grandmother of every pigeon, shark, mouse, dog, and, yes, human—swam unknowingly toward danger. She swam through the translucent underwater plants of the Cambrian, gently weaving between their thick seaweed-like stalks. She was hunting for coral larvae, the protein-rich offspring of the brainless animals populating the sea. Unbeknownst to her, she too was being hunted.
一只奇虾——一种一英尺长的节肢动物，头部长有两个带刺的爪子——藏在沙子里。奇虾是寒武纪的顶级捕食者，它正耐心地等待着不幸的生物进入它的攻击范围。
An Anomalocaris—a foot-long arthropod with two spiked claws sprouting from its head—lay hidden in the sand. Anomalocaris was the apex predator of the Cambrian, and it was waiting patiently for an unlucky creature to come within lunging distance.
我们的脊椎动物祖先应该注意到了这种陌生的气味和远处不规则形状的沙丘。但寒武纪海洋中总是有陌生的气味；它是一个微生物、植物、真菌和动物的动物园，每种动物都释放着自己独特的气味。而且总是有陌生形状的背景，无数物体（既有生命物体也有无生命物体）不断移动的肖像。所以她没怎么在意。
Our vertebrate ancestor would have noticed the unfamiliar smell and the irregular-shaped mound of sand in the distance. But there were always unfamiliar smells in the Cambrian ocean; it was a zoo of microbes, plants, fungi, and animals, each releasing their own unique portfolio of scents. And there was always a backdrop of unfamiliar shapes, an ever-moving portrait of countless objects, both living and inanimate. And so she thought nothing of it.
当她从寒武纪植物的安全地带出来时，节肢动物发现了她，并向前猛冲。几毫秒内，她的本能逃跑反应开始发挥作用。鱼奶奶的眼睛在她的周围发现了一个快速移动的物体，触发了硬连线的本能转身和冲刺反方向。这种逃跑反应的激活使她的大脑充满了去甲肾上腺素，引发了高度兴奋状态，使感官反应更加敏感，暂停所有恢复功能，并将能量重新分配到她的肌肉。在千钧一发之际，她逃脱了紧握的利爪，游走了。
As she emerged from the safety of the Cambrian plants, the arthropod spotted her and lurched forward. Within milliseconds, her reflexive escape response kicked in. Grandma Fish’s eyes detected a fast-moving object in her periphery, triggering a hardwired reflexive turn and dash in the opposite direction. The activation of this escape response flooded her brain with norepinephrine, triggering a state of high arousal, making sensory responses more sensitive, pausing all restorative functions, and reallocating energy to her muscles. In the nick of time, she escaped the clasping talons and swam away.
这种情况已经发生过数十亿次，这是一个永无止境的狩猎和逃跑、期待和恐惧的循环。但这一次不同——我们的脊椎动物祖先会记住那种危险节肢动物的气味；她会记得它的眼睛透过沙子窥视的景象。她不会再犯同样的错误。大约五亿年前，我们的祖先进化出了模式识别能力。
This has unfolded billions of times, a never-ending cycle of hunting and escaping, of anticipation and fear. But this time was different—our vertebrate ancestor would remember the smell of that dangerous arthropod; she would remember the sight of its eyes peeking through the sand. She wouldn’t make the same mistake again. Sometime around five hundred million years ago, our ancestor evolved pattern recognition.
识别气味是一个比你想象的更难的问题
The Harder-Than-You-Would-Think Problem of Recognizing a Smell
早期的两侧对称动物无法感知人类的嗅觉。尽管你很容易就能区分向日葵和鲑鱼的气味，但这实际上是一项非常复杂的智力壮举，是从第一批脊椎动物那里继承下来的。
Early bilaterians could not perceive what humans experience as smell. Despite how little effort it takes for you to distinguish the scent of a sunflower from that of a salmon, it is, in fact, a remarkably complicated intellectual feat, one inherited from the first vertebrates.
就像你今天的鼻子一样，早期脊椎动物的鼻孔中有成千上万的嗅觉神经元。七鳃鳗有大约五十种不同类型的嗅觉神经元，每种神经元都包含一个独特的嗅觉受体，可以对气味作出反应特定类型的分子。大多数气味不是由单个分子组成，而是由多个分子组成。当你回到家并闻到家人最美味的手撕猪肉的味道时，你的大脑并没有识别出手撕猪肉分子（没有这种东西）。相反，它正在识别一种由多种分子组成的特定混合物，这种混合物激活了嗅觉神经元的交响乐。任何给定的气味都由激活的嗅觉神经元模式表示。总之，气味识别只不过是模式识别。
Just as you have in your nose today, within the nostrils of early vertebrates were thousands of olfactory neurons. In the lamprey fish, there are about fifty different types of olfactory neurons, each type containing a unique olfactory receptor that responds to specific types of molecules. Most smells are made up not of a single molecule but of multiple molecules. When you come home and recognize the smell of your family’s best pulled pork, your brain isn’t recognizing the pulled-pork molecule (there is no such thing). Rather, it is recognizing a particular soup of many molecules that activates a symphony of olfactory neurons. Any given smell is represented by a pattern of activated olfactory neurons. In summary, smell recognition is nothing more than pattern recognition.
我们线虫类祖先识别世界的能力仅限于单个神经元的感觉机制。它可以通过激活单个感光神经元来识别光的存在，或者通过激活单个机械感觉神经元来识别触觉的存在。虽然这对于转向很有用，但这让人类痛苦不堪。模糊的外部世界画面。事实上，转向的奇妙之处在于，它使第一批两侧对称动物能够在不感知太多外部世界的情况下找到食物并躲避捕食者。
Our nematode-like ancestor’s ability to recognize the world was constrained to only the sensory machinery of individual neurons. It could recognize the presence of light by the activation of a single photosensitive neuron or the presence of touch from the activation of a single mechanosensory neuron. Although useful for steering, this rendered a painfully opaque picture of the outside world. Indeed, the brilliance of steering was that it enabled the first bilaterians to find food and avoid predators without perceiving much of anything about the world.
图 7.1：脊椎动物的鼻子内部
Figure 7.1: Inside the nose of a vertebrate
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
然而，你周围世界的大部分信息无法在单个激活的神经元中找到，而只能在激活的神经元模式中找到。你可以根据光子撞击视网膜的模式来区分汽车和房屋。你可以根据声波撞击内耳的模式来区分人的胡言乱语和豹子的吼叫。是的，你可以根据鼻子中激活的嗅觉神经元模式来区分玫瑰的气味和鸡肉的气味。数亿年来，动物被剥夺了这种技能，被困在感知的牢笼中。
However, most of the information about the world around you can’t be found in a single activated neuron but only in the pattern of activated neurons. You can distinguish a car from a house based on the pattern of photons hitting your retina. You can distinguish the ramblings of a person from the roar of a panther based on the pattern of sound waves hitting your inner ear. And, yes, you can distinguish the smell of a rose from the smell of chicken based on the pattern of olfactory neurons activated in your nose. For hundreds of millions of years, animals were deprived of this skill, stuck in a perceptual prison.
当你意识到盘子太热或针太锋利时，你正在以早期两侧对称动物的方式识别世界属性，激活单个神经元。然而，当你识别气味、面孔或声音时，你正在以超越早期两侧对称动物的方式识别世界上的事物；你正在使用一种后来在早期脊椎动物中出现的技能。
When you recognize that a plate is too hot or a needle too sharp, you are recognizing attributes of the world the way early bilaterians did, with the activations of individual neurons. However, when you recognize a smell, a face, or a sound, you are recognizing things in the world in a way that was beyond early bilaterians; you are using a skill that emerged later in early vertebrates.
早期脊椎动物可以利用大脑结构解码神经元模式来识别事物。这极大地扩展了动物的感知范围。在只有 50 种嗅觉神经元的小小马赛克中，存在着大量可以识别的不同模式。50 个细胞可以代表超过一百万亿种模式。*
Early vertebrates could recognize things using brain structures that decoded patterns of neurons. This dramatically expanded the scope of what animals could perceive. Within the small mosaic of only fifty types of olfactory neurons lived a universe of different patterns that could be recognized. Fifty cells can represent over one hundred trillion patterns.*
早期两侧对称动物如何认识世界 HOW EARLY BILATERIANS RECOGNIZED THINGS IN THE WORLD |
早期脊椎动物如何认识世界上的事物 HOW EARLY VERTEBRATES RECOGNIZED THINGS IN THE WORLD |
单个神经元检测特定事物 A single neuron detects a specific thing |
大脑解码激活的神经元的模式来识别特定的事物 Brain decodes the pattern of activated neurons to recognize a specific thing |
少数事物可以被识别 Small number of things can be recognized |
大量事物可以被识别 Large number of things can be recognized |
只有通过进化的修补才能识别新事物（需要新的传感机制） New things can be recognized only through evolutionary tinkering (new sensory machinery needed) |
无需进化调整，即可识别新事物，只需学习识别新模式即可（不需要新的感官机制） New things can be recognized without evolutionary tinkering but through learning to recognize a new pattern (no new sensory machinery needed) |
模式识别很难。当今生存的许多动物，即使经过了 5 亿年的进化，也从未获得这种能力——今天的线虫和扁虫没有表现出模式识别的迹象。
Pattern recognition is hard. Many animals alive today, even after another half billion years of evolution, never acquired this ability—the nematodes and flatworms of today show no evidence of pattern recognition.
脊椎动物的大脑需要解决两个计算难题才能识别模式。在图 7.2中，你可以看到三个虚构的气味模式：一个代表危险的捕食者，一个代表美味的食物，一个代表有吸引力的伴侣。从这个图中你也许可以看出为什么模式识别并不容易——这些模式虽然含义不同，但相互重叠。一个模式应该触发逃跑，另一个模式应该触发接近。这是模式识别的第一个问题，即辨别问题：如何将重叠的模式识别为不同的模式。
There were two computational challenges the vertebrate brain needed to solve to recognize patterns. In figure 7.2, you can see an example of three fictional smell patterns: one for a dangerous predator, one for yummy food, and one for an attractive mate. Perhaps you can see from this figure why pattern recognition won’t be easy—these patterns overlap with each other despite having different meanings. One should trigger escape and the others approach. This was the first problem of pattern recognition, that of discrimination: how to recognize overlapping patterns as distinct.
鱼第一次在小说面前感到恐惧每次遇到捕食者的气味，它都会记住特定的气味模式。但下次遇到同样的捕食者气味时，它不会激活完全相同的嗅觉神经元模式。分子的平衡永远不会完全相同——新节肢动物的年龄、性别、饮食或许多其他因素可能会有所不同，从而略微改变其气味。甚至周围环境的背景气味也可能不同，以略微不同的方式干扰。所有这些微小扰动的结果是，下一次遭遇将相似但不相同。在图 7.3中，你可以看到下一次遇到捕食者气味可能激活的嗅觉模式的三个例子。这是模式识别的第二个挑战：如何概括以前的模式以识别新的模式相似但不相同。
The first time a fish experiences fear in the presence of a novel predator smell, it will remember that specific smell pattern. But the next time the fish encounters that same predator smell, it won’t activate the exact same pattern of olfactory neurons. The balance of molecules will never be identical—the age of the new arthropod, or its sex, or its diet, or many other things might be different that could slightly alter its scent. Even the background smells from the surrounding environment might be different, interfering in slightly different ways. The result of all these minor perturbations is that the next encounter will be similar but not the same. In figure 7.3 you can see three examples of the olfactory patterns that the next encounter with the predator smell might activate. This is the second challenge of pattern recognition: how to generalize a previous pattern to recognize novel patterns that are similar but not the same.
图 7.2：歧视问题
Figure 7.2: The discrimination problem
图为 Max Bennett
Figure by Max Bennett
图 7.3：泛化问题
Figure 7.3: The generalization problem
图为 Max Bennett
Figure by Max Bennett
计算机如何识别模式
How Computers Recognize Patterns
你可以用脸部解锁 iPhone。要做到这一点，你的手机需要解决泛化和识别问题。你的 iPhone 需要能够区分你的脸部和其他人脸部，尽管脸部特征重叠（识别）。你的 iPhone 需要识别你的脸部，尽管阴影、角度、面部毛发等发生了变化（泛化）。显然，现代人工智能系统成功地解决了模式识别的这两个挑战。如何做到的？
You can unlock your iPhone with your face. Doing this requires your phone to solve the generalization and discrimination problems. Your iPhone needs to be able to tell the difference between your face and other people’s faces, despite the fact that faces have overlapping features (discrimination). And your iPhone needs to identify your face despite changes in shading, angle, facial hair, and more (generalization). Clearly, modern AI systems successfully navigate these two challenges of pattern recognition. How?
标准方法如下：创建一个如图 7.4所示的神经元网络，在网络的一端提供一个输入模式，该模式会流经多层神经元，直到在网络的另一端转换为输出。通过调整神经元之间连接的权重，可以让网络对其输入执行各种操作。如果可以编辑权重使其恰到好处，就可以获得一种算法来获取输入模式并在网络末端正确识别它。如果以一种方式编辑权重，它可以识别面部。如果以另一种方式编辑权重，它可以识别气味。
The standard approach is the following: Create a network of neurons like in figure 7.4 where you provide an input pattern on one side that flows through layers of neurons until they are transformed into an output on the other end of the network. By adjusting the weights of the connections between neurons, you can make the network perform a variety of operations on its input. If you can edit the weights to be just right, you can get an algorithm to take an input pattern and recognize it correctly at the end of the network. If you edit the weights one way, it can recognize faces. If you edit the weights a different way, it can recognize smells.
最难的部分是教会网络如何学习正确的权重。实现这一目标的最先进的机制是由 Geoffrey Hinton、David Rumelhart 和 Ronald Williams 在 20 世纪 80 年代推广的。他们的方法如下：如果你正在训练一个神经网络，将气味模式分类为鸡蛋气味或花朵气味，你会向它展示一堆气味模式，同时告诉网络每个模式是来自鸡蛋还是花朵（通过网络末端特定神经元的激活来衡量）。换句话说，你告诉网络正确的答案。然后，你将实际输出与期望输出进行比较，并朝着使实际输出更接近期望输出的方向调整整个网络的权重。如果你这样做很多次（比如，数百万次），网络最终就会学会准确识别模式——它可以识别鸡蛋和花朵的气味。他们将这种学习机制称为反向传播：它们将末端的错误传播回整个网络，计算每个突触的确切错误贡献，并相应地调整该突触。
The hard part is teaching the network how to learn the right weights. The state-of-the-art mechanism for doing this was popularized by Geoffrey Hinton, David Rumelhart, and Ronald Williams in the 1980s. Their method is as follows: If you were training a neural network to categorize smell patterns into egg smells or flower smells, you would show it a bunch of smell patterns and simultaneously tell the network whether each pattern is from an egg or a flower (as measured by the activation of a specific neuron at the end of the network). In other words, you tell the network the correct answer. You then compare the actual output with the desired output and nudge the weights across the entire network in the direction that makes the actual output closer to the desired output. If you do this many times (like, millions of times), the network eventually learns to accurately recognize patterns—it can identify smells of eggs and flowers. They called this learning mechanism backpropagation: they propagate the error at the end back throughout the entire network, calculate the exact error contribution of each synapse, and nudge that synapse accordingly.
上述类型的学习，即通过提供正确答案和示例来训练网络，称为监督学习（人类通过向网络提供正确答案来监督学习过程）。许多监督学习方法比这更复杂，但原理是一样的：提供正确答案，然后使用反向传播调整网络以更新权重，直到输入模式的分类足够准确。这种设计已被证明非常普遍，现已应用于图像识别、自然语言处理、语音识别和自动驾驶汽车。
The above type of learning, in which a network is trained by providing examples alongside the correct answer, is called supervised learning (a human has supervised the learning process by providing the network with the correct answers). Many supervised learning methods are more complex than this, but the principle is the same: the correct answers are provided, and networks are tweaked using backpropagation to update weights until the categorization of input patterns is sufficiently accurate. This design has proven to work so generally that it is now applied to image recognition, natural language processing, speech recognition, and self-driving cars.
但即使是反向传播的发明者之一杰弗里·辛顿 (Geoffrey Hinton) 也意识到，他的发明虽然有效，但却无法很好地反映大脑的实际工作方式。首先，大脑并不进行监督学习——当你得知一种气味是鸡蛋而另一种气味是草莓时，你并不会得到标记数据。甚至在孩子们还未学会鸡蛋和草莓这些词之前，他们就能清楚地识别出它们是不同的。其次，反向传播在生物学上是难以置信的。反向传播的工作原理是神奇地同时以恰当的量推动数百万个突触，从而使网络的输出朝正确的方向移动。大脑无法做到这一点。那么大脑是如何识别模式的呢？
But even one of the inventors of backpropagation, Geoffrey Hinton, realized that his creation, although effective, was a poor model of how the brain actually works. First, the brain does not do supervised learning—you are not given labeled data when you learn that one smell is an egg and another is a strawberry. Even before children learn the words egg and strawberry, they can clearly recognize that they are different. Second, backpropagation is biologically implausible. Backpropagation works by magically nudging millions of synapses simultaneously and in exactly the right amount to move the output of the network in the right direction. There is no conceivable way the brain could do this. So then how does the brain recognize patterns?
图 7.5：早期脊椎动物的皮层
Figure 7.5: The cortex of early vertebrates
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
皮质
The Cortex
鱼类的嗅觉神经元将其输出发送到大脑顶部的一个结构，称为皮层。像七鳃鳗和爬行动物这样的简单脊椎动物的皮层由一层薄薄的三层神经元。
Olfactory neurons of fish send their output to a structure at the top of the brain called the cortex. The cortex of simpler vertebrates, like the lamprey fish and reptiles, is made up of a thin sheet of three layers of neurons.
在第一个皮层中，出现了一种新的神经元形态，即锥体神经元，因其金字塔状的形状而得名。这些锥体神经元有数百个树突，通过数千个突触接收输入。这些是第一个为识别模式而设计的神经元。
In the first cortex evolved a new morphology of neuron, the pyramidal neuron, named after their pyramid-like shape. These pyramidal neurons have hundreds of dendrites and receive inputs across thousands of synapses. These were the first neurons designed for the purpose of recognizing patterns.
嗅觉神经元将信号发送到皮质的锥体神经元。这个将嗅觉输入皮质的网络具有两个有趣的特性。首先，它具有很大的维度扩展——少数嗅觉神经元连接到大量皮质神经元。其次，它们的连接很稀疏；一个特定的嗅觉细胞只会连接到这些皮质细胞的一个子集。这两个看似无害的布线特征可能会解决识别问题。
Olfactory neurons send their signals to the pyramidal neurons of the cortex. This network of olfactory input to the cortex has two interesting properties. First, there is a large dimensionality expansion—a small number of olfactory neurons connect to a much larger number of cortical neurons. Second, they connected sparsely; a given olfactory cell will connect to only a subset of these cortical cells. These two seemingly innocuous features of wiring may solve the discrimination problem.
图 7.6：扩展和稀疏性（也称为扩展重新编码）可以解决鉴别问题
Figure 7.6: Expansion and sparsity (also called expansion recoding) can solve the discrimination problem
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
使用图 7.6，您可以直观地了解为什么扩展和稀疏性可以实现这一点。即使捕食者气味和食物气味模式重叠，从所有激活的神经元获得输入的皮质神经元也会有所不同。因此，尽管输入重叠，但在皮质中激活的模式也会有所不同。此操作有时称为模式分离、去相关或正交化。
Using figure 7.6 you can intuit why expansion and sparsity achieve this. Even though the predator-smell and food-smell patterns are overlapping, the cortical neurons that get input from all the activated neurons will be different. As such, the pattern that gets activated in the cortex will be different despite the fact that the input is overlapping. This operation is sometimes called pattern separation, decorrelation, or orthogonalization.
神经科学家还发现了皮质如何解决泛化问题的线索。皮质的锥体细胞将其轴突发送回自身，并与数百到数千个附近的锥体细胞进行突触传递。这意味着，当一种气味模式激活了一组锥体神经元时，这组细胞就会通过赫布可塑性自动连接在一起。*下次出现某种模式时，即使它不完整，皮质也会重新激活完整的模式。这种技巧称为自动联想；皮质中的神经元会自动学习与自身的联想。这为泛化问题提供了一种解决方案——皮质可以识别相似但不相同的模式。
Neuroscientists have also found hints of how the cortex might solve the problem of generalization. Pyramidal cells of the cortex send their axons back onto themselves, synapsing on hundreds to thousands of other nearby pyramidal cells. This means that when a smell pattern activates a pattern of pyramidal neurons, this ensemble of cells gets automatically wired together through Hebbian plasticity.* The next time a pattern shows up, even if it is incomplete, the full pattern can be reactivated in the cortex. This trick is called auto-association; neurons in the cortex automatically learn associations with themselves. This offers a solution to the generalization problem—the cortex can recognize a pattern that is similar but not the same.
自动联想揭示了脊椎动物记忆与计算机记忆的一个重要区别。自动联想表明脊椎动物的大脑使用内容可寻址记忆——记忆是通过提供原始经验的子集来回忆的，这些子集会重新激活原始模式。如果我告诉你一个你以前听过的故事的开头，你可以回忆起剩下的部分；如果我给你看你的汽车的半幅图，你可以画出剩下的部分。然而，计算机使用寄存器可寻址内存——记忆只有拥有唯一的内存地址才能调用这些内容。如果丢失了地址，就丢失了内存。
Auto-association reveals an important way in which vertebrate memory differs from computer memory. Auto-association suggests that vertebrate brains use content-addressable memory—memories are recalled by providing subsets of the original experience, which reactivate the original pattern. If I tell you the beginning of a story you’ve heard before, you can recall the rest; if I show you half a picture of your car, you can draw the rest. However, computers use register-addressable memory—memories that can be recalled only if you have the unique memory address for them. If you lose the address, you lose the memory.
自动联想记忆不存在丢失记忆地址的难题，但它确实面临着另一种形式的遗忘问题。寄存器寻址存储器使计算机能够隔离信息的存储位置，确保新信息不会覆盖旧信息。相比之下，自动联想信息存储在共享的神经元群中，这使其面临意外覆盖旧记忆的风险。事实上，正如我们将看到的，这是使用神经元网络进行模式识别的一个基本挑战。
Auto-associative memory does not have this challenge of losing memory addresses, but it does struggle with a different form of forgetfulness. Register-addressable memory enables computers to segregate where information is stored, ensuring that new information does not overwrite old information. In contrast, auto-associative information is stored in a shared population of neurons, which exposes it to the risk of accidentally overwriting old memories. Indeed, as we will see, this is an essential challenge with pattern recognition using networks of neurons.
灾难性遗忘（或持续学习问题，第 2 部分）
Catastrophic Forgetting (or The Continual Learning Problem, Part 2)
1989 年，Neal Cohen 和 Michael McCloskey 试图教人工神经网络网络做数学运算。不是复杂的数学运算，只是加法。他们是约翰霍普金斯大学的神经科学家，他们都对神经网络如何存储和维护记忆感兴趣。当时，人工神经网络尚未进入主流，也尚未证明其许多实际用途——神经网络仍有待探索，以发现其缺失的功能和看不见的局限性。
In 1989, Neal Cohen and Michael McCloskey were trying to teach artificial neural networks to do math. Not complicated math, just addition. They were neuroscientists at Johns Hopkins, and they were both interested in how neural networks stored and maintained memories. This was before artificial neural networks had entered the mainstream, before they had proven their many practical uses—neural networks were still something to be probed for missing capabilities and unseen limitations.
Cohen 和 McCloskey 将数字转换成神经元模式，然后训练神经网络进行加法运算，即将两个输入数字（例如 1 和 3）转换为正确的输出数字（在本例中为 4）。他们首先教网络加 1（1+2、1+3、1+4 等），直到它熟练掌握为止。然后他们教同一个网络加 2（2+1、2+2、2+3 等），直到它也能熟练掌握为止。
Cohen and McCloskey converted numbers into patterns of neurons, then trained a neural network to do addition by transforming two input numbers (e.g., 1 and 3) into the correct output number (in this case, 4). They first taught the network to add ones (1+2, 1+3, 1+4, and so on) until it got good at it. Then they taught the same network to add twos (2+1, 2+2, 2+3, and so on) until it got good at this as well.
但随后他们注意到一个问题。在他们教会网络如何加二之后，网络却忘记了如何加一。当他们将错误传回网络并更新权重以教会它如何加二时，网络只是简单地覆盖了如何加一的记忆。它成功地学会了新任务，却以牺牲之前的任务为代价。
But then they noticed a problem. After they taught the network to add twos, it forgot how to add ones. When they propagated errors back through the network and updated the weights to teach it to add twos, the network had simply overridden the memories of how to add ones. It successfully learned the new task at the expense of the previous task.
Cohen 和 McCloskey 将人工神经网络的这种特性称为灾难性遗忘问题。这不是一个深奥的发现，而是神经网络普遍存在的、毁灭性的局限性：当你训练神经网络来识别新模式或执行新任务时，你可能会干扰网络先前学习到的模式。
Cohen and McCloskey referred to this property of artificial neural networks as the problem of catastrophic forgetting. This was not an esoteric finding but a ubiquitous and devastating limitation of neural networks: when you train a neural network to recognize a new pattern or perform a new task, you risk interfering with the network’s previously learned patterns.
现代人工智能系统如何克服这个问题？目前还没有。程序员只是通过在训练后冻结他们的人工智能系统来避免这个问题。我们不让人工智能系统按顺序学习东西；它们会一次性学习完所有东西，然后停止学习。
How do modern AI systems overcome this problem? Well, they don’t yet. Programmers merely avoid the problem by freezing their AI systems after they are trained. We don’t let AI systems learn things sequentially; they learn things all at once and then stop learning.
识别人脸、驾驶汽车或在放射图像中检测癌症的人工神经网络不会从新经验中不断学习。截至本书付印时，即使是 OpenAI 发布的著名聊天机器人 ChatGPT 也不会从与其交谈的数百万人中不断学习。它在被释放到世界上的那一刻也停止了学习。这些系统不允许学习新事物，因为它们有忘记旧事物（或学习错误事物）的风险。因此，现代人工智能系统在时间上是冻结的，它们的参数是锁定的；只有在从头开始重新训练时，人类会一丝不苟地监控它们在所有相关任务上的表现，它们才允许更新。
The artificial neural networks that recognize faces, drive cars, or detect cancer in radiology images do not learn continually from new experiences. As of this book going to print, even ChatGPT, the famous chatbot released by OpenAI, does not continually learn from the millions of people who speak to it. It too stopped learning the moment it was released into the world. These systems are not allowed to learn new things because of the risk that they will forget old things (or learn the wrong things). So modern AI systems are frozen in time, their parameters locked down; they are allowed to be updated only when retrained from scratch with humans meticulously monitoring their performance on all the relevant tasks.
当然，我们努力创造的类人人工智能并不是这样的。《杰森一家》里的罗西在你和她说话时会学习——你可以向她展示如何玩游戏，然后她就可以玩这个游戏，而不会忘记如何玩其他游戏。
The humanlike artificial intelligences we strive to create are, of course, not like this. Rosey from The Jetsons learned as you spoke to her—you could show her how to play a game and she could then play it without forgetting how to play other games.
虽然我们才刚刚开始探索如何使持续学习发挥作用，但动物的大脑已经这样做了很长时间。
While we are only just beginning to explore how to make continual learning work, animal brains have been doing so for a long time.
我们在第 4 章中看到，即使是早期的两侧对称动物也在不断学习；神经元之间的连接随着每次新体验而加强和减弱。但这些早期的两侧对称动物从未面临灾难性遗忘的问题，因为它们从一开始就从未学习过模式。如果仅使用单个感觉神经元来识别世界上的事物，那么这些感觉神经元和运动神经元之间的连接就可以加强和减弱而不会互相干扰。只有当知识以神经元模式表示时，例如在人工神经网络或脊椎动物的皮层中，学习新事物才有可能干扰对旧事物的记忆。
We saw in chapter 4 that even early bilaterians learned continually; the connections between neurons were strengthened and weakened with each new experience. But these early bilaterians never faced the problem of catastrophic forgetting because they never learned patterns in the first place. If things are recognized in the world using only individual sensory neurons, then the connection between these sensory neurons and motor neurons can be strengthened and weakened without interfering with each other. It is only when knowledge is represented in a pattern of neurons, like in artificial neural networks or in the cortex of vertebrates, that learning new things risks interfering with the memory of old things.
随着模式识别的发展，灾难性遗忘问题的解决方案也随之出现。事实上，即使是鱼也能避免灾难性遗忘。记忆力极佳。训练一条鱼通过一个小逃生舱逃出渔网，让它独自呆上一整年，然后再测试一次。在这段漫长的时间里，它的大脑会不断接收到模式流，不断学习识别新的气味、景象和声音。然而，当你一年后把鱼放回同一张渔网时，它会记得如何以几乎相同的速度和准确性逃脱前年。
As soon as pattern recognition evolved, so too did a solution to the problem of catastrophic forgetting. Indeed, even fish avoid catastrophic forgetting fantastically well. Train a fish to escape from a net through a small escape hatch, leave the fish alone for an entire year, and then test it again. During this long stretch of time, its brain will have received a constant stream of patterns, learning continually to recognize new smells, sights, and sounds. And yet, when you place the fish back in the same net an entire year later, it will remember how to get out with almost the same speed and accuracy as it did the year before.
关于脊椎动物的大脑如何做到这一点，有几种理论。一种理论认为，大脑皮层进行模式分离的能力使其免于灾难性遗忘的问题；通过在大脑皮层中分离传入的模式，模式本质上不太可能相互干扰。
There are several theories about how vertebrate brains do this. One theory is that the cortex’s ability to perform pattern separation shields it from the problem of catastrophic forgetting; by separating incoming patterns in the cortex, patterns are inherently unlikely to interfere with each other.
另一种理论认为，大脑皮层中的学习只在惊讶的时刻有选择地发生；只有当大脑皮层看到一种模式超过某种新颖性阈值时，突触的权重才会改变。这使得学习模式能够长时间保持稳定，因为学习只是有选择地发生。有证据表明，大脑皮层和丘脑之间的连接——这两种结构在早期脊椎动物中同时出现——总是在测量通过丘脑传入的感觉数据和大脑皮层中表示的模式之间的新颖性水平。如果匹配，则不允许学习，因此噪声输入不会干扰现有的学习模式。然而，如果存在不匹配——如果传入的模式足够新——那么这将触发神经调节剂释放的过程，从而触发大脑皮层突触连接的变化，使其现在能够学习这个新模式。
Another theory is that learning in the cortex selectively occurs only during moments of surprise; only when the cortex sees a pattern that passes some threshold of novelty are the weights of synapses allowed to change. This enables learned patterns to remain stable for long periods of time, as learning occurs only selectively. There is some evidence that the wiring between the cortex and the thalamus—both structures that emerged alongside each other in early vertebrates—are always measuring the level of novelty between incoming sensory data through the thalamus and the patterns represented in the cortex. If there is a match, then no learning is allowed, hence noisy inputs don’t interfere with existing learned patterns. However, if there is a mismatch—if an incoming pattern is sufficiently new—then this triggers a process of neuromodulator release, which triggers changes in synaptic connections in the cortex, enabling it to now learn this new pattern.
我们尚不清楚鱼类、爬行动物和两栖动物等简单的脊椎动物大脑如何克服灾难性遗忘的挑战。但下次你看到鱼时，你就会在它小小的软骨头部找到答案。
We do not yet understand exactly how simple vertebrate brains, like those of fish, reptiles, and amphibians, are capable of overcoming the challenges of catastrophic forgetting. But the next time you spot a fish, you will be in the presence of the answer, hidden in its small cartilaginous head.
不变性问题
The Invariance Problem
看看下面的两个物体。
Look at the two objects below.
当你观察每个物体时，眼睛后部的特定神经元模式会亮起来。眼球后部半毫米厚的微小膜眼球后部——视网膜——包含五种不同类型的一亿多个神经元。视网膜的每个区域接收来自视野不同位置的输入，每种类型的神经元对不同的颜色和对比度敏感。当你观察每个物体时，独特的神经元模式会激活一系列尖峰。就像构成气味模式的嗅觉神经元一样，视网膜中的神经元构成视觉模式；你的视力取决于你识别这些视觉模式的能力。
As you view each object, a specific pattern of neurons in the backs of your eyes light up. The minuscule half-millimeter-thick membrane in the back of the eye—the retina—contains over one hundred million neurons of five different types. Each region of the retina receives input from a different location of the visual field, and each type of neuron is sensitive to different colors and contrasts. As you view each object, a unique pattern of neurons activates a symphony of spikes. Like the olfactory neurons that make up a smell pattern, the neurons in the retina make up a visual pattern; your ability to see exists only in your ability to recognize these visual patterns.
图 7.7
Figure 7.7
在 SketchFab.com 上找到的免费 3D 对象
Free 3D objects found on SketchFab.com
视网膜中激活的神经元将信号发送到丘脑，丘脑随后将这些信号发送到处理视觉输入的皮质部分（视觉皮质）。视觉皮质解码和记忆视觉模式的方式与嗅觉皮质解码和记忆气味模式的方式相同。然而，视觉和嗅觉之间的相似之处仅此而已。
The activated neurons in the retina send their signals to the thalamus, which then sends these signals to the part of the cortex that processes visual input (the visual cortex). The visual cortex decodes and memorizes the visual pattern the same way the olfactory cortex decodes and memorizes smell patterns. This is, however, where the similarity between sight and smell ends.
看看下面的物体。你能辨认出哪些形状与第一张图片中的形状相同吗？
Look at the objects below. Can you identify which shapes are the same as the ones in the first picture?
图 7.8
Figure 7.8
在 SketchFab.com 上找到的免费 3D 对象
Free 3D objects found on SketchFab.com
你很容易就能看出图 7.8中的物体与图 7.7中的物体相同，这一事实令人难以置信。这取决于当你聚焦时，视网膜中激活的神经元可能完全不重叠，没有一个共享的神经元，但你仍然可以将它们识别为同一个物体。
The fact that it is so effortlessly obvious to you that the objects in figure 7.8 are the same as those in figure 7.7 is mind-blowing. Depending on where you focus, the activated neurons in your retina could be completely non-overlapping, with not a single shared neuron, and yet you could still identify them as the same object.
无论鸡蛋的旋转、距离或位置如何，鸡蛋气味激活的嗅觉神经元模式都是相同的。相同的分子在空气中扩散并激活相同的嗅觉神经元。但对于视觉等其他感官来说情况并非如此。
The pattern of olfactory neurons activated by the smell of an egg is the same no matter the rotation, distance, or location of the egg. The same molecules diffuse through the air and activate the same olfactory neurons. But this is not the case for other senses such as vision.
同一视觉对象可以根据其旋转、距离或视野中的位置激活不同的模式。这就产生了所谓的不变性问题：如何在输入存在很大差异的情况下将模式识别为相同。
The same visual object can activate different patterns depending on its rotation, distance, or location in your visual field. This creates what is called the invariance problem: how to recognize a pattern as the same despite large variances in its inputs.
我们所研究的关于皮层自联想的任何内容都无法令人满意地解释大脑如何如此轻松地做到这一点。我们描述的自联想网络无法从完全不同的角度识别你从未见过的物体。自联想网络会将这些视为不同的物体，因为输入神经元完全不同。
Nothing we have reviewed about auto-association in the cortex provides a satisfactory explanation for how the brain so effortlessly did this. The auto-associative networks we described cannot identify an object you have never seen before from completely different angles. An auto-associative network would treat these as different objects because the input neurons are completely different.
这不仅仅是视觉问题。当你能识别出孩子高音调的声音和成人低音调的声音说出的同一组单词时，你就是在解决不变性问题。由于声音的音调完全不同，你内耳中激活的神经元也不同，但你仍然可以分辨出它们是相同的单词。尽管感官输入存在巨大差异，但你的大脑还是以某种方式识别出了一种共同的模式。
This is not only a problem with vision. When you recognize the same set of words spoken by the high-pitched voice of a child and the low-pitched voice of an adult, you are solving the invariance problem. The neurons activated in your inner ear are different because the pitch of the sound is completely different, and yet you can still tell they are the same words. Your brain is somehow recognizing a common pattern despite huge variances in the sensory input.
1958 年，即科恩和麦克洛斯基发现灾难性遗忘问题的几十年前，约翰霍普金斯大学的另一个神经科学家团队正在探索模式识别的另一个方面。
In 1958, decades before Cohen and McCloskey discovered the problem of catastrophic forgetting, a different team of neuroscientists, also at Johns Hopkins, were exploring a different aspect of pattern recognition.
David Hubel 和 Torsten Wiesel 对猫进行麻醉，将电极插入猫的皮质，并记录神经元的活动。向猫展示不同的视觉刺激。他们在猫的视野中不同位置呈现点、线和各种形状。他们想知道皮层如何编码视觉输入。
David Hubel and Torsten Wiesel anesthetized cats, put electrodes into their cortices, and recorded the activity of neurons as they presented the cats with different visual stimuli. They presented dots, lines, and various shapes in different locations in the cats’ visual field. They wanted to know how the cortex encoded visual input.
在哺乳动物的大脑中（猫、老鼠、猴子、人类等），最先从眼睛接收输入的皮质部分称为V1（第一个视觉区域）。Hubel 和 Wiesel 发现 V1 中的单个神经元对它们所响应的内容具有惊人的选择性。一些神经元只被猫视野中特定位置的垂直线激活。其他神经元只被其他位置的水平线激活，还有一些神经元被不同位置的 45 度线激活。V1 的整个表面积构成了猫的整个视野的地图，单个神经元对每个位置的特定方向的线都有选择性。
In mammal brains (cats, rats, monkeys, humans, et cetera), the part of the cortex that first receives input from the eye is called V1 (the first visual area). Hubel and Wiesel discovered that individual neurons in V1 were surprisingly selective with what they respond to. Some neurons were activated only by vertical lines at a specific location in a cat’s visual field. Other neurons were activated only by horizonal lines at some other location, and still others by 45-degree lines at a different location. The entire surface area of V1 makes up a map of the cat’s full field of view, with individual neurons selective for lines of specific orientations at each location.
图片来自 Manassi 等人，2013 年。已获许可使用。
Figure from Manassi et al., 2013. Used with permission.
V1 将复杂的视觉输入模式分解为更简单的特征，如线条和边缘。从这里开始，视觉系统创建一个层次结构：V1 将其输出发送到附近的皮质区域 V2，然后 V2 发送信息到 V4，V4 再将信息发送到 IT 区域。
V1 decomposes the complex patterns of visual input into simpler features, like lines and edges. From here, the visual system creates a hierarchy: V1 sends its output to a nearby region of cortex called V2, which then sends information to an area called V4, which then sends information to an area called IT.
皮质层级结构中，水平越来越高的神经元对视觉刺激的复杂特征越来越敏感——V1 中的神经元主要由基本的边缘和线条激活，V2 和 V4 中的神经元对更复杂的形状和物体敏感，IT 中的神经元对复杂的整体物体（例如特定的面部）敏感。V1 中的神经元仅对视野特定区域的输入做出反应；相反，IT 中的神经元可以检测眼睛任何区域的物体。虽然 V1 将图片分解为简单特征，但随着视觉信息沿层级向上流动，它会被重新拼凑成整个物体。
Neurons at progressively higher levels of this cortical hierarchy become sensitive to progressively more sophisticated features of visual stimuli—neurons in V1 are primarily activated by basic edges and lines, neurons in V2 and V4 are sensitive to more complex shapes and objects, and neurons in IT are sensitive to complex whole objects such as specific faces. A neuron in V1 responds only to input in a specific region of one’s visual field; in contrast, a neuron in IT can detect objects across any region of the eye. While V1 decomposes pictures into simple features, as visual information flows up the hierarchy, it is pieced back together into whole objects.
20 世纪 70 年代末，在 Hubel 和 Wiesel 的初步工作完成二十多年后，一位名叫 Kunihiko Fukushima 的计算机科学家试图让计算机识别图片中的物体。尽管他尽了最大努力，他无法让标准神经网络（如本章前面描述的神经网络）成功做到这一点；即使物体的位置、旋转或大小发生微小变化，也会激活完全不同的神经元组，这使得网络无法将不同的模式推广到同一物体上——这里的正方形会被错误地认为与那里的正方形不同。他在不变性问题上遇到了麻烦。他知道，大脑以某种方式解决了这个问题。
In the late 1970s, well over twenty years after Hubel and Wiesel’s initial work, a computer scientist by the name of Kunihiko Fukushima was trying to get computers to recognize objects in pictures. Despite his best attempts, he couldn’t get standard neural networks, like those depicted earlier in the chapter, to successfully do it; even small changes in the location, rotation, or size of an object activated entirely different sets of neurons, which blinded networks to generalizing different patterns to the same object—a square over here would be incorrectly perceived as different from the same square over there. He had stumbled on the invariance problem. And he knew that somehow, brains solved it.
福岛此前四年一直在一个由多位神经生理学家组成的研究小组工作，因此他熟悉 Hubel 和 Wiesel 的工作。Hubel 和 Wiesel 发现了两件事。首先，哺乳动物的视觉处理是分层的，较低层次的接受域较小，可识别较简单的特征，较高层次的接受域较大，可识别较复杂的物体。其次，在层次结构的给定层次上，神经元都对类似特征敏感，只是位置不同。例如，V1 的一个区域会在一个位置寻找线条，另一个区域会在另一个位置寻找线条，但它们都在寻找线条。
Fukushima had spent the prior four years working in a research group that included several neurophysiologists, and so he was familiar with the work of Hubel and Wiesel. Hubel and Wiesel had discovered two things. First, visual processing in mammals was hierarchical, with lower levels having smaller receptive fields and recognizing simpler features, and higher levels having larger receptive fields and recognizing more complex objects. Second, at a given level of the hierarchy, neurons were all sensitive to similar features, just in different places. For example, one area of V1 would look for lines at one location, and another area would look for lines for another location, but they were all looking for lines.
福岛预感到这两项发现是大脑如何解决不变性问题的线索，因此福岛发明了一种新的人工神经网络架构，旨在捕捉这两个想法由 Hubel 和 Wiesel 发现。他的架构背离了拍摄照片并将其放入完全连接的神经网络的标准方法。他的架构首先将输入图片分解为多个特征图，就像 V1 似乎做的那样。每个特征图都是一个网格，表示特征（例如垂直线或水平线）在输入图片中的位置。这个过程称为卷积，因此这个名字适用于福岛发明的网络类型：卷积神经网络。*
Fukushima had a hunch that these two findings were clues as to how brains solved the invariance problem, and so Fukushima invented a new architecture of artificial neural networks, one designed to capture these two ideas discovered by Hubel and Wiesel. His architecture departed from the standard approach of taking a picture and throwing it into a fully connected neural network. His architecture first decomposed input pictures into multiple feature maps, like V1 seemed to do. Each feature map was a grid that signaled the location of a feature—such as vertical or horizontal lines—within the input picture. This process is called a convolution, hence the name applied to the type of network that Fukushima had invented: convolutional neural networks.*
在这些特征图识别出某些特征后，它们的输出被压缩并传递到另一组可以组合的特征图将它们分解成更广泛图片区域的更高级特征，将线条和边缘合并成更复杂的物体。所有这些都是为了模拟哺乳动物大脑皮层的视觉处理而设计的。令人惊讶的是，它确实奏效了。
After these feature maps identified certain features, their output was compressed and passed to another set of feature maps that could combine them into higher-level features across a wider area of the picture, merging lines and edges into more complex objects. All this was designed to be analogous to the visual processing of the mammalian cortex. And, amazingly, it worked.
图 7.10：卷积神经网络
Figure 7.10: A convolutional neural network
图为 Max Bennett 设计的。狗狗照片来自 Oscar Sutton（在 Unsplash 上购买）。
Figure designed by Max Bennett. The dog photo is from Oscar Sutton (purchased on Unsplash).
从自动驾驶汽车到检测放射图像中肿瘤的算法，大多数使用计算机视觉的现代人工智能系统都使用福岛的卷积神经网络。人工智能曾经是盲人，但现在可以看见东西，这一天赋可以追溯到五十多年前探测猫神经元的发明。
Most modern AI systems that use computer vision, from your self-driving car to the algorithms that detect tumors in radiology images use Fukushima’s convolutional neural networks. AI was blind, but now can see, a gift that can be traced all the way back to probing cat neurons over fifty years ago.
福岛的卷积神经网络的妙处在于它强加了一种巧妙的“归纳偏差”。归纳偏差是人工智能系统根据其设计方式做出的假设。卷积神经网络的设计基于平移不变性的假设，即一个位置的给定特征应与不同位置的相同特征相同。这是我们视觉世界的一个不容置疑的事实：同一事物可以存在于不同的地方，而事物本身并无不同。因此，福岛没有试图让任意的神经元网络学习视觉世界的这一事实（这需要太多的时间和数据），而是直接将这一规则编码到网络架构中。
The brilliance of Fukushima’s convolutional neural network is that it imposes a clever “inductive bias.” An inductive bias is an assumption made by an AI system by virtue of how it is designed. Convolutional neural networks are designed with the assumption of translational invariance, that a given feature in one location should be treated the same as that same feature but in a different location. This is an impregnable fact of our visual world: the same thing can exist in different places without the thing being different. And so, instead of trying to get an arbitrary web of neurons to learn this fact about the visual world, which would require too much time and data, Fukushima simply encoded this rule directly into the architecture of the network.
尽管卷积神经网络 (CNN) 的灵感来源于大脑，但实际上它并不能很好地模拟大脑识别视觉模式的方式。首先，视觉处理并不像最初想象的那样具有层次性；输入经常跳过层级并分支到多个层级同时。其次，CNN 施加了平移约束，但它们本身并不理解 3D 物体的旋转，因此在旋转时无法很好地识别物体。*第三，现代 CNN 仍然建立在监督和反向传播的基础上——神奇的是它可以同时更新许多连接——而大脑皮层似乎可以在没有监督和反向传播的情况下识别物体。
Despite being inspired by the brain, convolutional neural networks (CNNs) are, in fact, a poor approximation of how brains recognize visual patterns. First, visual processing isn’t as hierarchical as originally thought; input frequently skips levels and branches out to multiple levels simultaneously. Second, CNNs impose the constraint of translation, but they don’t inherently understand rotations of 3D objects, and thus don’t do a great job recognizing objects when rotated.* Third, modern CNNs are still founded on supervision and backpropagation—with its magical simultaneous updating of many connections—while the cortex seems to recognize objects without supervision and without backpropagation.
第四，也许是最重要的一点，CNN 受到哺乳动物视觉皮层的启发，该皮层比鱼类简单的视觉皮层复杂得多；然而，鱼脑——缺乏任何明显的层次结构或哺乳动物皮层的其他花哨功能——仍然能够解决不变性问题。
And fourth, and perhaps most important, CNNs were inspired by the mammal visual cortex, which is much more complex than the simpler visual cortex of fish; and yet the fish brain—lacking any obvious hierarchy or the other bells and whistles of the mammalian cortex—is still eminately capable of solving the invariance problem.
2022 年，罗彻斯特理工学院的比较心理学家 Caroline DeLong 训练金鱼轻敲图片来获取食物。她向金鱼展示了两张图片。每当金鱼轻敲一张青蛙图片时，她就给金鱼喂食。金鱼很快学会了只要看到青蛙图片，就会游到图片前。德朗随后改变了实验。她向金鱼展示了同一只青蛙的图片，但从金鱼从未见过的新角度。如果金鱼无法从不同角度识别同一物体，它们就会像对待其他照片一样对待这张照片。然而，令人惊讶的是，金鱼游到了新的青蛙图片前，尽管角度不同，但显然能够立即认出青蛙，就像你在几页前认出 3D 物体一样。
In 2022, the comparative psychologist Caroline DeLong at Rochester Institute of Technology trained goldfish to tap pictures to get food. She presented the goldfish with two pictures. Whenever the fish tapped specifically a picture of a frog, she gave the fish food. Fish quickly learned to swim right up to the frog picture whenever it was presented. DeLong then changed the experiment. She presented the picture of the same frog but from new angles that the fish had never seen before. If fish were unable to recognize the same object from different angles, they would have treated this like any other photo. And yet, amazingly, the fish swam right up to the new frog picture, clearly able to immediately recognize the frog despite the new angle, just as you recognized the 3D objects a few pages ago.
鱼脑是如何做到这一点的还不得而知。虽然自联想捕捉到了模式识别在大脑皮层中如何运作的一些原理，但显然，即使是鱼的大脑皮层也在做一些更复杂的事情。一些人认为，脊椎动物大脑解决不变性问题的能力并非来自哺乳动物独特的大脑皮层结构，而是来自大脑皮层和丘脑之间复杂的相互作用，这种相互作用自第一批脊椎动物出现以来就一直存在。也许丘脑——大脑中心的一个球形结构——就像一个三维黑板一样运作，大脑皮层提供最初的感官输入，丘脑智能地将这种感觉信息传送到皮质的其他区域，共同将二维输入渲染成完整的三维物体，从而能够灵活地识别旋转和平移的物体。
How the fish brain does this is not understood. While auto-association captures some principles of how pattern recognition works in the cortex, clearly even the cortex of fish is doing something far more sophisticated. Some theorize that the vertebrate brain’s ability to solve the invariance problem derives not from the unique cortical structures in mammals, but from the complex interactions between the cortex and the thalamus, interactions that have been present since the first vertebrates. Perhaps the thalamus—a ball-shaped structure at the center of the brain—operates like a three-dimensional blackboard, with the cortex providing initial sensory input and the thalamus intelligently routing this sensory information around other areas of the cortex, together rendering full 3D objects from 2D input, thereby flexibly able to recognize rotated and translated objects.
也许 CNN 给我们最好的启示不是它们试图模仿的特定假设的成功——比如平移不变性——而是假设本身的成功。事实上，虽然 CNN 可能无法准确捕捉大脑的工作方式，但它们揭示了良好归纳偏差的力量。在模式识别中，良好的假设使学习变得快速而高效。脊椎动物的皮层肯定有这样的归纳偏差，我们只是不知道它是什么。
Perhaps the best lesson from CNNs is not the success of the specific assumptions they attempt to emulate—such as translational invariance—but the success of assumptions themselves. Indeed, while CNNs may not capture exactly how the brain works, they reveal the power of a good inductive bias. In pattern recognition, it is good assumptions that make learning fast and efficient. The vertebrate cortex surely has such an inductive bias, we just don’t know what it is.
在某些方面，微小的鱼脑超越了我们一些最好的计算机视觉系统。CNN 需要大量数据才能理解旋转和 3D 物体的变化，但鱼似乎可以一次性识别 3D 物体的新角度。
In some ways, the tiny fish brain surpasses some of our best computer-vision systems. CNNs require incredible amounts of data to understand changes in rotations and 3D objects, but a fish seems to recognize new angles of a 3D object in one shot.
在寒武纪的掠食性军备竞赛中，进化从为动物配备新的感觉神经元来检测特定事物，转变为为动物配备一般机制来识别任何事物。
In the predatory arms race of the Cambrian, evolution shifted from arming animals with new sensory neurons for detecting specific things to arming animals with general mechanisms for recognizing anything.
凭借这种新的模式识别能力，脊椎动物的感觉器官的复杂性激增，并迅速发展成为现代形式。鼻子进化为检测化学物质；内耳进化为检测声音频率；眼睛进化为检测视觉。脊椎动物熟悉的感觉器官和熟悉的大脑的共同进化并非巧合——它们彼此促进对方的增长和复杂性。大脑模式识别的每一次渐进式改进都扩大了拥有更详细的感觉器官所带来的好处；而感觉器官细节的每一次渐进式改进都扩大了更复杂的模式识别所带来的好处。
With this new ability of pattern recognition, vertebrate sensory organs exploded with complexity, quickly flowering into their modern form. Noses evolved to detect chemicals; inner ears evolved to detect frequencies of sound; eyes evolved to detect sights. The coevolution of the familiar sensory organs and the familiar brain of vertebrates is not a coincidence—they each facilitated the other’s growth and complexity. Each incremental improvement to the brain’s pattern recognition expanded the benefits to be gained by having more detailed sensory organs; and each incremental improvement in the detail of sensory organs expanded the benefits to be gained by more sophisticated pattern recognition.
在大脑中，其结果就是脊椎动物皮层，它以某种方式无需监督就能识别模式，以某种方式准确地区分重叠模式并将模式推广到新的体验，以某种方式不断学习模式而不会遭受灾难性的遗忘，并且以某种方式识别模式尽管其输入存在很大的差异。
In the brain, the result was the vertebrate cortex, which somehow recognizes patterns without supervision, somehow accurately discriminates overlapping patterns and generalizes patterns to new experiences, somehow continually learns patterns without suffering from catastrophic forgetting, and somehow recognizes patterns despite large variances in its input.
模式识别和感觉器官的精细化反过来也发现自己与强化学习本身处于一个反馈循环中。模式识别和强化学习在进化过程中同时发展也并非巧合。大脑学习对世界事物做出任意反应的能力越强，从识别更多世界事物中获得的好处就越大。大脑能够识别的独特物体和位置越多，它就能学会采取的独特行动就越多。因此，皮层、基底神经节和感觉器官一起进化，它们都来自强化学习的相同机制。
The elaboration of pattern recognition and sensory organs, in turn, also found themselves in a feedback loop with reinforcement learning itself. It is also not a coincidence that pattern recognition and reinforcement learning evolved simultaneously in evolution. The greater the brain’s ability to learn arbitrary actions in response to things in the world, the greater the benefit to be gained from recognizing more things in the world. The more unique objects and places a brain can recognize, the more unique actions it can learn to take. And so the cortex, basal ganglia, and sensory organs evolved together, all emerging from the same machinations of reinforcement learning.
8
8
生活为何变得奇妙
Why Life Got Curious
在 TD-Gammon 取得成功后，研究人员开始将 Sutton 的时间差分学习应用到各种不同的游戏中。一个接一个，之前“无法解决”的游戏被这些算法成功击败；TD 学习算法最终在 Pinball、Star Gunner、Robotank、Road Runner、Pong 和 Space Invaders 等视频游戏中超越了人类水平的表现。然而，有一款 Atari 游戏令人费解地遥不可及：蒙特祖玛的复仇。
IN THE AFTERMATH of the success of TD-Gammon, researchers began applying Sutton’s temporal difference learning to all kinds of different games. And one by one, games that had previously been “unsolvable” were successfully beaten by these algorithms; TD learning algorithms eventually surpassed human-level performance in video games like Pinball, Star Gunner, Robotank, Road Runner, Pong, and Space Invaders. And yet there was one Atari game that was perplexingly out of reach: Montezuma’s Revenge.
在《蒙特祖玛的复仇》中，玩家从一间布满障碍的房间开始。每个方向都有一间房间，每个房间都有自己的障碍。没有任何迹象或线索表明哪个方向是正确的方向。当你找到通往远处隐藏房间中隐藏门的路时，你将获得第一个奖励。这使得游戏对于强化学习系统来说特别困难：第一个奖励在游戏后期出现，因此没有早期提示应该强化或惩罚哪些行为。然而，人类当然以某种方式赢得了这个游戏。
In Montezuma’s Revenge, you start in a room filled with obstacles. In each direction is another room, each with its own obstacles. There is no sign or clue as to which direction is the right way to go. The first reward is earned when you find your way to a hidden door in a faraway hidden room. This makes the game particularly hard for reinforcement learning systems: the first reward occurs so late in the game that there is no early nudging of what behavior should be reinforced or punished. And yet somehow, of course, humans beat this game.
直到 2018 年，一种算法才最终完成蒙特祖玛复仇的第一关。这种由谷歌 DeepMind 开发的新算法通过添加 Sutton 最初的 TD 学习算法中缺少的熟悉元素来实现这一壮举：好奇心。
It wasn’t until 2018 when an algorithm was developed that finally completed level one of Montezuma’s Revenge. This new algorithm, developed by Google’s DeepMind, accomplished this feat by adding something familiar that was missing from Sutton’s original TD learning algorithm: curiosity.
萨顿一直知道，任何强化学习系统都会遇到一个问题，即所谓的“开发-探索困境”。要使试错学习发挥作用，智能体需要进行大量的试验来学习。这意味着强化学习不能通过仅利用它们预测的行为就会获得奖励；它还必须探索新的行为。
Sutton had always known that a problem with any reinforcement learning system is something called the exploitation-exploration dilemma. For trial-and-error learning to work, agents need to, well, have lots of trials from which to learn. This means that reinforcement learning can’t work by just exploiting behaviors they predict lead to rewards; it must also explore new behaviors.
换句话说，强化学习需要两个对立过程——一个针对之前得到强化的行为（利用），另一个针对新行为（探索）。从定义上讲，这些选择是相互对立的。利用总是会驱使行为朝着已知的奖励方向发展，而探索总是会驱使行为朝着未知的方向发展。
In other words, reinforcement learning requires two opponent processes—one for behaviors that were previously reinforced (exploitation) and the other for behaviors that are new (exploration). These choices are, by definition, opposing each other. Exploitation will always drive behavior toward known rewards, and exploration will always drive toward what is unknown.
在早期的 TD 学习算法中，这种权衡的实现方式很粗糙：这些 AI 系统会自发地（比如说，5% 的时间）做一些完全随机的事情。如果你玩的是一款受限游戏，只有有限的几个下一步动作，这种方法还可以，但在像《蒙特祖玛的复仇》这样的游戏中，这种方法就很糟糕，因为你可以去的方向和地点实际上是无限的。
In early TD learning algorithms, this trade-off was implemented in a crude way: these AI systems spontaneously—say, 5 percent of the time—did something totally random. This worked okay if you were playing a constrained game with only so many next moves, but it worked terribly in a game like Montezuma’s Revenge, where there were practically an infinite number of directions and places you could go.
还有一种解决开发-探索困境的替代方法，这种方法既非常简单，又令人耳目一新。这种方法是让人工智能系统明确地表现出好奇心，奖励它们探索新地方和做新事物，让惊喜本身得到强化。新奇事物越大，探索的冲动就越大。当玩《蒙特祖玛的复仇》的人工智能系统被赋予这种探索新事物的内在动机时，它们的行为非常不同——实际上更像人类玩家。它们有动力探索区域，进入新房间，并在整个地图上扩张。但它们不是通过随机行动进行探索，而是有意识地探索；它们特别想去新地方和做新事物。
There is an alternative approach to tackling the exploitation-exploration dilemma, one that is both beautifully simple and refreshingly familiar. The approach is to make AI systems explicitly curious, to reward them for exploring new places and doing new things, to make surprise itself reinforcing. The greater the novelty, the larger the compulsion to explore it. When AI systems playing Montezuma’s Revenge were given this intrinsic motivation to explore new things, they behaved very differently—indeed, more like a human player. They became motivated to explore areas, go to new rooms, and expand throughout the map. But instead of exploring through random actions, they explored deliberately; they specifically wanted to go to new places and to do new things.
尽管在通过第一关之前没有明确的奖励，但这些人工智能系统不需要任何外部奖励来探索。它们自我激励。找到通往新房间的路本身就很有价值。有了好奇心的武装，这些模型突然开始取得进展，最终通过了第一关。
Even though there are no explicit rewards until you get past all the rooms in level one, these AI systems didn’t need any external rewards to explore. They were motivated on their own. Simply finding their way to a new room was valuable in and of itself. Armed with curiosity, suddenly these models started making progress, and they eventually beat level one.
好奇心在强化学习算法中的重要性表明，通过强化学习而设计的大脑（例如早期脊椎动物的大脑）也应该表现出好奇心。事实上，有证据表明，早期脊椎动物最先表现出好奇心。从鱼到老鼠再到猴子，所有脊椎动物都表现出好奇心对人类婴儿。在脊椎动物中，惊喜本身会触发多巴胺的释放，即使有没有“真正的”奖励。然而，大多数无脊椎动物并不表现出好奇心；只有最先进的无脊椎动物，如昆虫和头足类动物表现出好奇心，这是独立进化而来的技巧，在早期的两侧对称动物中并不存在。
The importance of curiosity in reinforcement learning algorithms suggests that a brain designed to learn through reinforcement, such as the brain of early vertebrates, should also exhibit curiosity. And indeed, evidence suggests that it was early vertebrates who first became curious. Curiosity is seen across all vertebrates, from fish to mice to monkeys to human infants. In vertebrates, surprise itself triggers the release of dopamine, even if there is no “real” reward. And yet, most invertebrates do not exhibit curiosity; only the most advanced invertebrates, such as insects and cephalopods, show curiosity, a trick that evolved independently and wasn’t present in early bilaterians.
好奇心的出现和机制有助于解释赌博，这是脊椎动物行为中一种非理性的怪异现象。赌徒违反了桑代克效应定律——尽管预期回报为负数，他们仍继续赌钱。
The emergence and mechanisms of curiosity help explain gambling, which is an irrational oddity of vertebrate behavior. Gamblers violate Thorndike’s law of effect—they continue to gamble their money away despite the fact that the expected reward is negative.
BF Skinner 是第一个意识到老鼠会赌博的人。让老鼠执着地推杠杆以获得食物的最好方法不是每次按下杠杆时都释放食物颗粒，而是让杠杆随机释放食物颗粒。这种可变比率强化会让老鼠发疯，它们无休止地推杠杆，似乎痴迷于看看再推一次是否能产生颗粒。即使释放的颗粒总数总体相同，这种可变比率强化也比固定比率强化导致的杠杆推力要多得多。鱼类也会表现出这种效应。
B. F. Skinner was the first to realize that rats will gamble. The best way to get a rat to obsessively push a lever for food is not to have the lever release food pellets every time it is pressed; instead, it is to have the lever randomly release food pellets. Such variable-ratio reinforcement makes rats go crazy; they endlessly push the lever, seemingly obsessed with seeing if just one more push might be the one to produce a pellet. Even when the total number of pellets released are identical overall, such variable-ratio reinforcement leads to far more lever pushes than fixed-ratio reinforcement. Fish exhibit this effect as well.
对此的一个解释是，当某事令人惊讶时，脊椎动物会得到额外的强化。为了让动物感到好奇，我们进化出寻找令人惊讶和新奇事物来强化自己，这驱使我们去追求和探索它们。这意味着，即使某项活动的奖励是负面的，如果它是新奇的，我们还是会去追求它。
One explanation for this is that vertebrates get an extra boost of reinforcement when something is surprising. To make animals curious, we evolved to find surprising and novel things reinforcing, which drives us to pursue and explore them. This means that even if the reward of an activity is negative, if it is novel, we might pursue it anyway.
赌博游戏就是为了利用这一点而精心设计的。在赌博游戏中，你赢钱的几率不会是 0%（这会导致你不玩）；你有 48% 的赢钱几率，这个几率高到足以让你赢钱，这个几率不确定到让你在赢钱时感到意外（让你多巴胺激增），这个几率低到足以让赌场从长远来看榨干你。
Games of gambling are carefully designed to exploit this. In games of gambling, you don’t have a 0 percent chance of winning (which would lead you not to play); you have a 48 percent chance of winning, high enough to make it possible, uncertain enough to make it surprising when you win (giving you a dopamine boost), and low enough so that the casino will, in the long run, suck you dry.
我们的 Facebook 和 Instagram 动态也利用了这一点。每次滚动都会出现一条新帖子，滚动几次后，会随机出现一些有趣的内容。即使你可能不想使用 Instagram，就像赌徒不想赌博或吸毒者不想再吸毒一样，这种行为也会在潜意识中得到强化，越来越难以停止。
Our Facebook and Instagram feeds exploit this as well. With each scroll, there is a new post, and randomly, after some number of scrolls, something interesting shows up. Even though you might not want to use Instagram, the same way gamblers don’t want to gamble or drug addicts don’t want to use anymore, the behavior is subconsciously reinforced, making it harder and harder to stop.
赌博和社交信息流通过侵入我们五亿年来对惊喜的偏好来发挥作用，产生了一种进化来不及解释的适应不良的边缘情况。
Gambling and social feeds work by hacking into our five-hundred-million-year-old preference for surprise, producing a maladaptive edge case that evolution has not had time to account for.
好奇心和强化学习共同进化，因为好奇心是强化学习发挥作用的必要条件。随着新发现的识别模式、记住地点和根据过去的奖励和惩罚灵活改变行为的能力，第一批脊椎动物迎来了一个新的机会：学习本身首次成为一项极其有价值的活动。脊椎动物识别的模式越多，记住的地点越多，生存能力就越强。她尝试的新事物越多，她就越有可能学会自己的行为与相应结果之间的正确对应关系。5 亿年前，好奇心首次出现在我们类似鱼类的祖先的小脑中。
Curiosity and reinforcement learning coevolved because curiosity is a requirement for reinforcement learning to work. With the newfound ability to recognize patterns, remember places, and flexibly change behavior based on past rewards and punishments, the first vertebrates were presented with a new opportunity: for the first time, learning became, in and of itself, an extremely valuable activity. The more patterns a vertebrate recognized and the more places she remembered, the better she would survive. And the more new things she tried, the more likely she was to learn the correct contingencies between her actions and their corresponding outcomes. And so it was 500 million years ago in the tiny brain of our fish-like ancestors when curiosity first emerged.
9
9
世界第一模特
The First Model of the World
你有没有试过在黑暗中穿过你的家？我猜不是故意的，但也许是在停电或半夜漫步去卫生间的时候。如果你曾经尝试过，你可能会意识到（这并不奇怪）这很难做到。当你走出卧室，走向走廊尽头时，你很容易错误地预测走廊的长度或卫生间门的确切位置。你可能会撞到脚趾。
HAVE YOU EVER tried to navigate through your home in the dark? I’m guessing not on purpose, but perhaps during a power outage or a midnight stroll to the bathroom. If you have ever tried this, you probably had the (not very surprising) realization that it is hard to do. As you step out of your bedroom and walk toward the end of the hall, you are prone to mis-predicting the length of the hallway or the exact location of the bathroom door. You might stub a toe.
但你也会注意到，尽管你看不见，但你对走廊尽头的位置还是有合理的预感，对你在迷宫般的家中的位置有一定的直觉。你可能走错一两步，但你的直觉仍然是一个有效的指引。这一点的非凡之处不在于它很难，而在于它完全可以实现。
But you would also notice that, despite your blindness, you have a reasonable hunch about where the end of the hallway is, some intuition about where you are in the labyrinth of your home. You might be off by a step or two, but your intuition nonetheless proves an effective guide. What is remarkable about this is not that it is hard, but that it is achievable at all.
你之所以能做到这一点，是因为你的大脑已经构建了你家的空间地图。你的大脑有一个你家的内部模型，因此，当你移动时，你的大脑可以自行更新你在地图上的位置。这种构建外部世界内部模型的能力是从第一批脊椎动物的大脑中继承下来的。
The reason you can do this is that your brain has built a spatial map of your home. Your brain has an internal model of your home, and thus, as you move, your brain can update your position in this map on its own. This trick, the ability to construct an internal model of the external world, was inherited from the brains of first vertebrates.
鱼类地图
The Maps of Fish
这种在黑暗中找到浴室的测试也可以在鱼身上进行。好吧，不是浴室部分，而是在没有视觉指引的情况下记住位置的一般测试。把一条鱼放在一个空鱼缸里，鱼缸里有 25 个相同的容器。把食物藏在一个容器。鱼会探索鱼缸，随机检查每个容器，直到它偶然发现食物。现在把鱼从鱼缸里拿出来，把食物放回同一个容器里，再把鱼放回鱼缸里。这样做几次，鱼就会学会快速地游动直接放入装有食物的容器中。
This same find-your-way-to-the-bathroom-in-the-dark test can be done in fish. Well, not the bathroom part, but the general test of remembering a location without a visual guide. Put a fish in an empty tank with a grid of twenty-five identical containers throughout the tank. Hide food in one of the containers. The fish will explore the tank, randomly inspecting each container until it stumbles on the food. Now take the fish out of the tank, put food back in the same container, and put the fish back into the tank. Do this a few times, and the fish will learn to quickly dart directly to the container with the food.
鱼并没有学习“看到这个物体时总是向左转”这样的固定规则——无论它们最初被放在鱼缸的什么位置，它们都会游到正确的位置。它们也没有学习“朝着食物的图像或气味游”这样的固定规则；即使你没有将任何食物放回容器中，鱼也会回到正确的容器中。换句话说，即使每个容器都完全相同（因为它们都没有食物），鱼仍然能够正确识别哪个容器当前放置在之前装有食物的确切位置。
Fish are not learning some fixed rule of Always turn left when I see this object—they navigate to the correct location no matter where in the tank they are initially placed. And they are not learning some fixed rule of Swim toward this image or smell of food; fish will go back to the correct container even if you don’t put any food back in the container. In other words, even if every container is exactly identical because none of them have any food at all, fish still correctly identify which container is currently placed at the exact location that previously contained food.
唯一能确定哪个容器之前盛放食物的线索是鱼缸壁，壁上有标记来标明特定的侧面。因此，鱼只能根据容器相对于鱼缸侧面地标的位置来识别正确的容器。鱼能够做到这一点的唯一方法是在它们的脑海中构建一个空间地图——一个世界的内部模型。
The only clue as to which container previously held the food was the walls of the tank itself, which had markings to designate specific sides. Thus, fish somehow identified the correct container based solely on the container’s location relative to the landmarks on the side of the tank. The only way fish could have accomplished this is by building a spatial map—an internal model of the world—in their minds.
学习空间地图的能力在脊椎动物中随处可见。鱼类、爬行动物、老鼠、猴子和人类都有这种能力。然而，像线虫这样的简单两侧对称动物却无法学习这样的空间地图——它们无法记住一件事物相对于另一件事物。
The ability to learn a spatial map is seen across vertebrates. Fish, reptiles, mice, monkeys, and humans all do this. And yet simple bilaterians like nematodes are incapable of learning such a spatial map—they cannot remember the location of one thing relative to another thing.
甚至许多高级无脊椎动物，如蜜蜂和蚂蚁，也无法解决空间问题。请考虑以下对蚂蚁的研究。蚂蚁从巢穴到食物形成一条路径，从食物源到巢穴形成另一条路径。它们带着拾到的食物返回巢穴，然后空手而归，去寻找更多的食物。假设你把一只蚂蚁带回巢穴，放在离开巢穴的路上。这只蚂蚁显然想回到巢穴，但现在被放在了一个它只离开过巢穴的地方。如果蚂蚁有一个内部空间模型，它会意识到回家的最快路线就是转身朝相反的方向走。这就是鱼会做的事情。但如果蚂蚁只学会了一系列动作（在提示 X 处右转，在提示 Y 处左转），那么它就会乖乖地重新开始循环。事实上，蚂蚁去再次完成整个循环。蚂蚁通过遵循一组何时转向何处的规则来导航，而不是通过构建空间地图。
Even many advanced invertebrates such as bees and ants are unable to solve spatial tasks. Consider the following ant study. Ants form a path from a nest to food and another path from the food source to their nest. They return to the nest with scavenged food, then leave again empty-handed to acquire more food. Suppose you took one of the ants on its way back to the nest and placed it on the path that is leaving the nest. This ant clearly wants to go back to the nest, but is now placed in a location from which it has only ever navigated away from the nest. If the ant had an internal model of space, it would realize that the fastest way home is to just turn around and go in the opposite direction. That is what a fish would do. But if instead the ant learned only a series of movements (turn right at cue X, turn left at cue Y), then it would just dutifully begin the loop all over again. Indeed, ants go through the entire loop again. Ants navigate by following a set of rules of when to turn where, not by constructing a map of space.
你的内心指南针
Your Inner Compass
这是您可以对自己进行的另一项测试：坐在其中一张旋转椅上，闭上眼睛，让某人转动椅子，然后在睁开眼睛之前猜出您面向房间的哪个方向。您的准确率会令人吃惊。您的大脑是如何做到这一点的？
Here’s another test you can do on yourself: Sit in one of those swivel chairs, close your eyes, ask someone to turn the chair, and then guess what direction of the room you are facing before opening your eyes. You will be amazingly accurate. How did your brain do this?
内耳深处是半规管，即充满液体的小管。这些管内排列着漂浮在液体中的感觉神经元，只要检测到运动就会激活。半规管分为三个环路，一个环路面向前方，一个环路面向侧面，一个环路面向上方。只有当你在特定维度上移动时，每个半规管中的液体才会移动。因此，激活的感觉细胞群会发出头部运动方向的信号。这创造了一种独特的感觉——前庭感觉。这就是为什么如果你在旋转中会感到头晕的原因椅子——最终，这会过度激活这些感觉细胞，当你停止转动时，它们仍然活跃，即使你没有旋转也会错误地发出旋转信号。
Deep in your inner ear are semicircular canals, small tubes filled with fluid. These canals are lined with sensory neurons that float in this fluid and activate whenever they detect movement. The semicircular canals are organized in three loops, one for facing forward, one for facing sideways, and one for facing upward. The fluid in each of these canals moves only when you move in that specific dimension. Thus, the ensemble of activated sensory cells signal the direction of head movement. This creates a unique sense—the vestibular sense. This is why you get dizzy if you are spun in a chair continuously—eventually, this overactivates these sensory cells and when you stop turning, they are still active, incorrectly signaling rotation even when you aren’t rotating.
Figure 9.1: The vestibular sense of fish emerges from the uniquely vertebrate semicircular canals
图片由 Carlyn Iverson / Science Source 提供。经许可使用。
Image by Carlyn Iverson / Science Source. Used with permission.
半规管的进化起源于早期脊椎动物，与强化学习和构建空间地图的能力同时出现。现代鱼类的内耳也有同样的结构，这使它们能够识别自己何时运动以及运动的程度。
The evolutionary origin of semicircular canals was in early vertebrates, emerging at the same time as reinforcement learning and the ability to construct spatial maps. Modern fish have the same structure in their inner ears, and it enables them to identify when and by how much they are moving.
前庭感知是构建空间地图的必要特征。动物需要能够分辨出向它游来的东西和它向着别处游来的东西之间的区别。在每种情况下，视觉线索都是相同的（都显示物体正在靠近），但就空间运动而言，每种线索的含义截然不同。前庭系统帮助鱼分辨出区别：如果它开始向某个物体游去，前庭系统就会检测到这种加速度。相反，如果一个物体开始向它移动，就不会发生这种激活。
The vestibular sense is a necessary feature of building a spatial map. An animal needs to be able to tell the difference between something swimming toward it and it swimming toward something. In each case, the visual cues are the same (both show an object coming closer), but each means very different things in terms of movement through space. The vestibular system helps the fish tell the difference: If it starts swimming toward an object, the vestibular system will detect this acceleration. In contrast, if an object starts moving toward it, no such activation will occur.
在鱼类和大鼠等不同物种的脊椎动物的后脑中，存在着所谓的“头部方向神经元”，只有当动物朝向某个方向。这些细胞整合视觉和前庭输入，形成神经指南针。脊椎动物的大脑从一开始就进化为模拟和导航三维空间。
In the hindbrain of vertebrates, in species as diverse as fish and rats, are what are called “head-direction neurons” that fire only when an animal is facing a certain direction. These cells integrate visual and vestibular input to create a neural compass. The vertebrate brain evolved, from its very beginnings, to model and navigate three dimensional space.
但是，如果鱼的后脑构建了动物自身方向的指南针，那么外部空间的模型又在哪里构建呢？脊椎动物的大脑在哪里存储关于事物相对于其他事物的位置的信息？
But if the hindbrain of fish constructs a compass of an animal’s own direction, where is the model of external space constructed? Where does the vertebrate brain store information about the locations of things relative to other things?
内侧皮质（又称海马体）
The Medial Cortex (aka the Hippocampus)
这早期脊椎动物的皮层有三个亚区：侧皮层，腹侧皮层和内侧皮层。外侧皮层是早期脊椎动物识别气味的区域，后来进化为早期哺乳动物的嗅觉皮层。腹侧皮层是早期脊椎动物学习视觉和声音模式的区域，后来进化为早期哺乳动物的杏仁核。但折叠到大脑中间的是第三个区域，即内侧皮层。
The cortex of early vertebrates had three subareas: lateral cortex, ventral cortex, and medial cortex. The lateral cortex is the area where early vertebrates recognized smells and that would later evolve into the olfactory cortex in early mammals. The ventral cortex is the area where early vertebrates learned patterns of sights and sounds and that would later evolve into the amygdala in early mammals. But folded into the middle of the brain was the third area, the medial cortex.
图 9.2：早期脊椎动物的皮层
Figure 9.2: The cortex of early vertebrates
原创艺术由 Mesa Schumacher 创作
Original art by Mesa Schumacher
内侧皮质是皮质的一部分，后来成为哺乳动物的海马体。如果你记录鱼在游动时海马体中的神经元，你会发现有些神经元只有当鱼在空间中的特定位置时才会激活，有些神经元只有当鱼在水箱边缘时才会激活，有些神经元只有当鱼面向特定方向。视觉、前庭和头部方向信号传播到内侧皮层，在那里它们混合在一起并转换成空间地图。
The medial cortex is the part of cortex that later became the hippocampus in mammals. If you record neurons in the hippocampus of fish as they navigate around, you will find some neurons that activate only when the fish are at a specific location in space, others only when the fish are at a border of a tank, and others only when the fish are facing specific directions. Visual, vestibular, and head-direction signals propagate to the medial cortex, where they are all mixed together and converted into a spatial map.
事实上，如果你损坏了鱼的海马体，它虽然可以学会朝提示游去或远离提示，但它会失去记忆位置的能力。这些鱼无法利用远处的地标来判断正确的方向在迷宫中转弯；无法在开放的环境中导航到特定位置找不到空间去获取食物；当给定不同的起始位置。
Indeed, if you damage the hippocampus of a fish, it can learn to swim toward or away from cues, but it loses the ability to remember locations. These fish fail to use distant landmarks to figure out the right direction to turn in a maze; fail to navigate to specific locations in an open space to get food; and fail to figure out how to escape a simple room when given different starting locations.
海马体的功能和结构在许多脊椎动物谱系中都得到了保留。在人类和大鼠中，海马体含有位置细胞，这些神经元只有当动物处于开放迷宫中的特定位置时才会激活。蜥蜴、大鼠和人类的海马体受损同样会损害空间导航。
The function and structure of the hippocampus has been conserved across many lineages of vertebrates. In humans and rats, the hippocampus contains place cells, which are neurons that activate only when an animal is in a specific location in an open maze. Damage to the hippocampus in lizards, rats, and humans similarly impairs spatial navigation.
显然，早期脊椎动物的三层皮层计算能力远远超出了简单的自动联想。它不仅似乎能够在旋转和缩放发生很大变化的情况下识别物体（解决了不变性问题），而且似乎还能够构建空间的内部模型。推测：也许皮层在旋转变化的情况下识别物体的能力与其建模空间的能力有关。也许皮层被调整为建模 3D事物——无论这些事物是物体还是空间地图。
Clearly the three-layered cortex of early vertebrates performed computations far beyond simple auto-association. Not only is it also seemingly capable of recognizing objects despite large changes in rotation and scale (solving the invariance problem), but it is also seemingly capable of constructing an internal model of space. To speculate: Perhaps the ability of the cortex to recognize objects despite changes in rotation and its ability to model space are related. Perhaps the cortex is tuned to model 3D things—whether those things are objects or spatial maps.
早期脊椎动物大脑中空间地图的进化标志着许多第一次。这是生命史上十亿年中生物第一次能够识别自己所在位置。不难想象这将带来的优势。虽然大多数无脊椎动物会四处走动并执行反射性运动反应，但早期脊椎动物可以记住节肢动物倾向于躲藏的地方、如何回到安全的地方以及充满食物的角落和缝隙的位置。
The evolution of spatial maps in the minds of early vertebrates marked numerous firsts. It was the first time in the billion-year history of life that an organism could recognize where it was. It is not hard to envision the advantage this would have offered. While most invertebrates steered around and executed reflexive motor responses, early vertebrates could remember the places where arthropods tended to hide, how to get back to safety, and the locations of nooks and crannies filled with food.
这也是大脑第一次将自我与世界区分开来。为了在空间地图上追踪自己的位置，动物需要能够区分“有东西向我游来”和“我向某物游来”之间的区别。
It was also the first time a brain differentiated the self from the world. To track one’s location in a map of space, an animal needs to be able to tell the difference between “something swimming toward me” and “me swimming toward something.”
最重要的是，这是大脑首次构建内部模型——外部世界的表征。这种模型的最初用途很可能很普通：它使大脑能够识别空间中的任意位置，并计算从任何起始位置到给定目标位置的正确方向。但这种内部模型的构建为大脑进化的下一个突破奠定了基础。最初作为记忆位置的技巧，后来变得更加重要。
And most important, it was the first time that a brain constructed an internal model—a representation of the external world. The initial use of this model was, in all likelihood, pedestrian: it enabled brains to recognize arbitrary locations in space and to compute the correct direction to a given target location from any starting location. But the construction of this internal model laid the foundation for the next breakthrough in brain evolution. What began as a trick for remembering locations would go on to become much more.
突破 2 总结：强化
Summary of Breakthrough #2: Reinforcing
大约五亿年前，我们的祖先从简单的蠕虫状两侧对称动物过渡到鱼类状脊椎动物。这些早期脊椎动物的大脑中出现了许多新的大脑结构和能力，其中大部分可以理解为是突破 #2 的促成和诞生：强化学习。这些包括
Our ancestors from around five hundred million years ago transitioned from simple wormlike bilaterians to fishlike vertebrates. Many new brain structures and abilities emerged in these early vertebrate brains, most of which can be understood as enabling and emerging from breakthrough #2: reinforcement learning. These include
- 多巴胺成为一种时间差异学习信号，有助于解决时间信用分配问题，使动物能够通过反复试验进行学习。
- Dopamine became a temporal difference learning signal, which helped solve the temporal credit assignment problem and enabled animals to learn through trial and error.
- 基底神经节作为一个演员-评论系统出现，使动物能够通过预测未来的奖励来产生这种多巴胺信号，并使用这种多巴胺信号来强化和惩罚行为。
- The basal ganglia emerged as an actor-critic system, enabling animals to generate this dopamine signal by predicting future rewards and to use this dopamine signal to reinforce and punish behaviors.
- 好奇心成为强化学习发挥作用（解决探索-利用困境）的必要部分。
- Curiosity emerged as a necessary part of making reinforcement learning work (solving the exploration-exploitation dilemma).
- 大脑皮层作为一个自联想网络出现，使得模式识别成为可能。
- The cortex emerged as an auto-associative network, making pattern recognition possible.
- 精确时间感知的出现使得动物能够通过反复试验，不仅学会做什么，还能学会何时做。
- The perception of precise timing emerged, enabling animals to learn, through trial and error, not only what to do but when to do it.
- 三维空间的感知出现了（在海马体和其他结构中），使动物能够识别它们在哪里，并记住事物相对于其他事物的位置。
- The perception of three-dimensional space emerged (in the hippocampus and other structures), enabling animals to recognize where they were and remember the location of things relative to other things.
早期脊椎动物之所以能够进行强化学习，是因为效价和联想学习的机制已经在早期的两侧对称动物中进化出来了。强化学习是基于更简单的好坏效价信号进行的。从概念上讲，脊椎动物的大脑建立在两侧对称动物更古老的操控系统之上。没有操控，就没有试错的起点，也没有衡量强化或取消强化的基础。
Reinforcement learning in early vertebrates was possible only because the mechanisms of valence and associative learning had already evolved in early bilaterians. Reinforcement learning is bootstrapped on simpler valence signals of good and bad. Conceptually, the vertebrate brain is built on top of the more ancient steering system of bilaterians. Without steering, there is no starting point for trial and error, no foundation on which to measure what to reinforce or un-reinforce.
图 9.3
Figure 9.3
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
操纵两侧对称动物使得后来的脊椎动物能够通过反复试验来学习。而脊椎动物的反复试验又使得随后更加令人困惑和具有里程碑意义的突破成为可能。早期哺乳动物首先想出了如何进行另一种反复试验的方法：不是通过实践而是通过想象来学习。
Steering bilaterians made it possible for later vertebrates to learn through trial and error. And trial and error in vertebrates, in turn, made it possible for the even more perplexing and monumental breakthrough that would follow. It was early mammals who first figured out how to engage in a different flavor of trial and error: learning not by doing but by imagining.
突破＃3
Breakthrough #3
模拟和第一批哺乳动物
Simulating and the First Mammals
2亿年前的大脑
Your brain 200 million years ago
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
10
10
神经黑暗时代
The Neural Dark Ages
从 4.2 亿年前到 3.75亿年前，海洋中充满了各种形状和大小的掠食性鱼类，种类也越来越多。如今的鲨鱼和黄貂鱼很常见。20 英尺长的盾皮鱼，头上有铠甲，牙齿厚实，能咬碎骨头，位于这条食物链的顶端。
FROM 420 TO 375 million years ago, oceans became filled with progressively more diverse predatory fish of many shapes and sizes. What would have resembled the sharks and stingrays of today were common sightings. Twenty-foot-long placoderms, fish with armored head plates and thick bone-crushing teeth, found themselves at the top of this food chain.
节肢动物和其他无脊椎动物被归类到不同的生态位。有些变小了。有些进化出了更厚的外壳。有些甚至从早期脊椎动物那里获得了启发，并通过变得更聪明而幸存下来——正是在这个时期，头足类动物出现了，它们是今天的鱿鱼和章鱼的祖先。在鱼类大规模捕食的巨大生存压力下，头足类动物变得非常聪明，它们形成了一条独立的谱系，其大脑的工作原理与我们截然不同。
Arthropods and other invertebrates were relegated to various niches. Some got smaller. Some evolved thicker shells. Some even took a cue from early vertebrates and survived by getting smarter—it was during this period that the cephalopods emerged, the ancestors of today’s squids and octopuses. Under severe pressure to survive their mass hunting by fish, cephalopods became impressively intelligent down an independent lineage with brains that work very differently than our own.
无脊椎动物最激进的生存策略是一起逃离海洋。节肢动物是第一批走出海洋、在陆地上繁衍生息的动物，它们被无情的掠食者赶出了家园。它们在海岸边稀疏生长的小型无叶陆生植物中找到了喘息之机。
The most radical invertebrate survival strategy was to escape the sea all together. The arthropods, driven from their homeland by relentless predation, were the first animals to walk out of the oceans and populate the land. They found respite among the small leafless land-faring plants that had sparsely sprouted along the seashores.
距今 4.2 亿至 3.75 亿年前的时期被称为泥盆纪，正是在这一时期，陆生植物首次进化出叶子，以便更好地吸收阳光，并进化出种子以便传播，这两者都使植物能够繁殖到以前不适宜居住的地区。类似于今天的树木的植物首先发展起来，长出粗壮的根，为附近的节肢动物创造稳定的土壤。在泥盆纪早期，陆地上的植物不超过三十厘米高，但到了泥盆纪末期，它们是三十米高。直到此时，随着陆地植物遍布地球表面，我们的星球才开始从上方呈现绿色。
The period between 420 to 375 million years ago is called the Devonian period, and it was here when land plants first evolved leaves for better absorption of sunlight and seeds for spreading, both of which enabled plants to propagate to previously inhospitable areas. Plants that resembled today’s trees first developed, growing thick roots and creating stable soils for nearby arthropods to live. In the early Devonian period, plants on land were no more than thirty centimeters tall, but by the end of it, they were thirty meters tall. It was only at this point that our planet began to appear green from above, as land plants spread across Earth’s surface.
虽然节肢动物在海里的生活令人恐惧，但在陆地上却如天堂般美好。节肢动物发展出新的技巧来满足陆地生活的需求，分化成类似于今天的蜘蛛和昆虫的生物。不幸的是，正如我们在今天的气候变化问题中看到的那样，地球的生物圈对那些快速和不可持续地繁殖的生物是无情的。最初作为节肢动物难民的小绿洲，最终变成了植物生命的过度狂欢，引发了全球灭绝事件，导致近一半的生命灭绝。
While life for arthropods was horrifying in the sea, it was heavenly on land. Arthropods developed new tricks to meet the needs of life on land, diversifying into what resembled today’s spiders and insects. Unfortunately, as we have seen with today’s problem of climate change, Earth’s biosphere is unforgiving to those who proliferate rapidly and unsustainably. What began as a small oasis for arthropod refugees eventually became an overextended orgy of plant life, triggering a global extinction event that would eradicate close to half of all life.
两位伟大人物的死亡故事
A Story of Two Great Deaths
历史总是重演。
History repeats itself.
十五亿年前，蓝藻的激增使地球充满二氧化碳，氧气污染了地球。十多亿年后，陆地上植物的激增似乎也犯下了同样的罪行。
One and a half billion years ago, the explosion of cyanobacteria suffocated the Earth with carbon dioxide and polluted it with oxygen. Over a billion years later, the explosion of plants on land seems to have committed a similar crime.
植物向内陆迁移的速度太快，以至于进化无法通过更多产生二氧化碳的动物的扩张来适应和重新平衡二氧化碳水平。二氧化碳水平急剧下降，导致气候变冷。海洋结冰，逐渐变得不适合生命生存。这是晚泥盆世灭绝事件是这个时代的第一次大规模灭绝。关于灭绝的原因，存在着许多不同的理论；有人认为，灭绝的原因不是植物过度繁殖，而是其他自然灾害。无论如何，我们的祖先正是从这场悲剧的冰冷坟墓中走出海洋的。
The inland march of plants was too rapid for evolution to accommodate and rebalance carbon dioxide levels through the expansion of more CO2-producing animals. Carbon dioxide levels plummeted, which caused the climate to cool. The oceans froze over and gradually became inhospitable to life. This was the Late Devonian Extinction, the first great death of this era. There are competing theories of what caused it; some argue that it was not overproliferation of plants but some other natural disaster. In any case, it was from the icy graves of this tragedy that our ancestors emerged from the sea.
灭绝事件为小生态位转变为主导策略创造了机会。在晚泥盆纪灭绝之前，我们的祖先已经找到了这样的生态位。大多数鱼都远离海岸，以避免致命的搁浅，即潮水退去时鱼被困在陆地上的情况。虽然搁浅的风险使追逐变得危险，但在靠近海岸的地方可以找到巨大的营养奖品：温暖的泥土水坑里充满了小昆虫和植被。
Extinction events create opportunities for small niches to transform into dominant strategies. Before the Late Devonian Extinction, our ancestors had found such a niche. Most fish stayed far away from the shore to avoid deadly beaching, a situation where a fish becomes stuck on land as tides recede. Although the risk of beaching made it dangerous to pursue, there was a big nutritional prize to be found close to the shore: the warm earthy puddles were full of small insects and vegetation.
我们的祖先是第一批进化出在水生昆虫进化出了一对肺，扩大了鳃，使它们能够从水和空气中提取氧气。因此，我们的祖先会用鳍在水中游泳，也会在陆地上涉水短距离，从一个水坑游到另一个水坑寻找昆虫。
Our ancestors were the first fish to evolve the ability to survive out of water. They developed a pair of lungs that augmented their gills, enabling them to extract oxygen from both water and air. And so our ancestors would use their fins both for swimming in water and for wading themselves short distances on land, traveling from puddle to puddle in search of insects.
当晚泥盆纪灭绝事件开始使海洋结冰时，我们呼吸空气、在陆地上行走的祖先是少数幸存的温水鱼之一。随着温水中的食物供应开始减少，我们的祖先更多地生活在内陆水坑中。他们失去了鳃（因此失去了在水下呼吸的能力），蹼鳍被手指状的手和脚取代。他们成为第一批四足动物（“tetra”代表“四”，“pods”代表“脚”），与蝾螈等现代两栖动物最为相似。
When the Late Devonian Extinction Event began to freeze over the oceans, our air-breathing and land-walking ancestors were one of the few warm-water fish to survive. As the food supply in warm waters began to die, our ancestors spent more of their time living in the inland puddles. They lost their gills (and thus their ability to breath underwater), and their webbed fins gave way to fingered hands and feet. They became the first tetrapods (tetra for “four” and pods for “feet”), most closely resembling a modern amphibian such as a salamander.
有一支四足动物的进化谱系很幸运地生活在地球上仍有这些较温暖的水坑的地区，它们维持了这种生活方式数亿年——它们成为了今天的两栖动物。另一支谱系放弃了濒临灭绝的海岸，向内陆深处游荡寻找食物。这就是羊膜动物谱系——这种生物进化出了产下革质卵的能力，这些卵可以在水外生存。
One evolutionary lineage of tetrapods, who were lucky enough to live in parts of the Earth that still supported these warmer puddles, would maintain this lifestyle for hundreds of millions of years—they would become the amphibians of today. Another lineage abandoned the dying shores and wandered farther inland in search of food. This was the lineage of amniotes—the creatures that developed the ability to lay leathery eggs that could survive out of the water.
第一批羊膜动物可能与今天的蜥蜴最为相似。羊膜动物发现内陆生态系统食物丰富——昆虫和植物随处可见，可供它们享用。最终，泥盆纪冰河时代消退，羊膜动物传播并多样化到地球的各个角落。石炭纪和二叠纪时期（合计从 3.5 亿年前持续到 2.5 亿年前）见证了陆地上羊膜动物的爆发式增长。
The first amniotes probably best resembled a lizard of today. Amniotes found an inland ecosystem abundant with food—insects and plants were everywhere for the feasting. Eventually, the Devonian ice age faded and amniotes spread and diversified to all corners of the Earth. The Carboniferous and Permian eras, which collectively lasted from 350 million years ago to 250 million years ago, saw an explosion of amniotes on land.
生活在陆地上给羊膜动物带来了独特的挑战，而它们的鱼类表亲从未面临过这样的挑战。其中一个挑战就是温度波动。在海洋深处，昼夜循环和季节循环只会造成温和的温度变化。相比之下，海面的温度可能会剧烈波动。羊膜动物和鱼一样，都是冷血动物——它们调节体温的唯一策略就是搬到更温暖的地方。
Living on land presented unique challenges to the amniotes that their fish cousins never faced. One such challenge was temperature fluctuations. Cycles of the day and season create only muted temperature changes deep in the oceans. In contrast, temperatures can fluctuate dramatically on the surface. Amniotes, like fish, were cold-blooded—their only strategy for regulating their body temperature was to physically relocate to warmer places.
羊膜动物谱系之一是爬行动物，它们最终分化为恐龙、蜥蜴、蛇和海龟。这些爬行动物大多与通过在夜间保持静止来应对每日温度波动。温度太低，它们的肌肉和新陈代谢无法正常运作，所以它们干脆停止活动。爬行动物一生中有三分之一的时间处于停止活动状态，这一事实提供了一个机会——能够在夜间捕猎的生物将收获一顿令人难以置信的盛宴，一动不动的蜥蜴。
One amniote lineage were the reptiles, who would eventually diversify into dinosaurs, lizards, snakes, and turtles. Most of these reptiles dealt with daily temperature fluctuations by becoming immobile at night. Temperatures were too low for their muscles and metabolisms to function properly, so they simply shut down. The fact that reptiles were shut down for a third of their lives presented an opportunity—creatures that could hunt in the night would reap an incredible feast of motionless lizards.
羊膜动物的另一个谱系是我们的祖先：兽孔目动物。兽孔目动物与当时的爬行动物有一个重要区别：它们进化出了温血特性。兽孔目动物是第一种进化出利用能量产生体内热量能力的脊椎动物。*这是一场赌博。它们需要更多的食物才能生存，但作为回报，它们有能力随时捕猎，包括它们的爬行动物表亲一动不动地躺在寒冷的夜晚——这是二叠纪盛宴上的一道轻松盛宴。
The other lineage of amniotes were our ancestors: the therapsids. The therapsids differed from reptiles at the time in one important way: they developed warm-bloodedness. Therapsids were the first vertebrates to evolve the ability to use energy to generate their own internal heat.* This was a gamble. They would require far more food to survive, but in return they had the ability to hunt at any time, including the cold nights when their reptile cousins lay immobile—an easy feast offered on a Permian platter.
在二叠纪时期，陆地上到处都是可食用的爬行动物和节肢动物，这种冒险获得了回报。在 3 亿至 2.5 亿年前的时期，兽孔目动物成为最成功的陆地动物。它们长到现代老虎的大小，并开始长毛以进一步保持体温。这些兽孔目动物看起来就像大型多毛蜥蜴。
In the Permian era, when the land was full of edible reptiles and arthropods, this gamble paid off. During the period from 300 to 250 million years ago, therapsids became the most successful land animals. They grew to the size of a modern tiger and began to grow hair to further maintain their heat. These therapsids would have looked like large hairy lizards.
或许你已经从地球生命的进化史中看到了一个趋势：所有统治都会结束。地球上的兽孔类统治也不例外：大约 2.5 亿年前发生的二叠纪-三叠纪大规模灭绝事件是地球历史上最致命的灭绝事件。这是这个时代的第二次大灭绝。这次灭绝事件是最严重的，也许是最神秘的。在五百万到一千万年内，96% 的海洋生物死亡，70% 的陆地生物死亡。关于造成这一现象的原因仍然存在争议——理论包括小行星、火山爆发和产生甲烷的微生物。一些人认为这不是单一的原因，而是多次不幸事件的完美风暴。不管原因是什么，我们都知道其影响。
Perhaps you can already see a trend emerging from the evolutionary history of life on Earth: all reigns come to an end. The therapsid reign on Earth was no different: the Permian-Triassic mass extinction event, which occurred around 250 million years ago, was the deadliest of all extinction events in Earth’s history. It was the second great death in this era. This extinction event was the most severe and perhaps the most enigmatic. Within five to ten million years, 96 percent of all marine life died, and 70 percent of all land life died. There is still controversy over what caused this—theories include asteroids, volcanic explosions, and methane-producing microbes. Some suggest that it was no single reason but rather a perfect storm of multiple unlucky occurrences. Regardless of the cause, we know the effects.
大型兽孔目动物几乎完全灭绝。最初促成它们崛起的温血动物的赌博，也是它们灭绝的原因衰落。在食物短缺的时期，兽孔类动物由于需要大量卡路里而首先灭绝。爬行动物及其相对稀少的食物更适合经受住这场风暴。
The large therapsids went almost entirely extinct. The gamble of warm-bloodedness that originally facilitated their rise was also the cause of their downfall. During a period of reduced access to food, the therapsids, with their need for huge amounts of calories, died first. The reptiles and their comparatively scant diets were much better suited to weather this storm.
图 10.1：第一个兽孔类
Figure 10.1: The first therapsid
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
在大约五百万年的时间里，生命只在世界的极小区域存活下来。唯一幸存下来的兽孔目动物是小型食草动物，比如穴居犬齿兽。犬齿兽最初进化出在地下穴居的生态位，以躲避统治世界的体型更大、更具掠食性的兽孔目动物。随着食物供应的消失和所有大型动物的灭绝，这些小型犬齿兽成为二叠纪-三叠纪灭绝事件后出现的少数幸存兽孔目动物之一。
For about five million years, life survived only in tiny pockets of the world. The only therapsids that survived were the small plant-eating ones, such as the burrowing cynodonts. The cynodonts originally evolved into the niche of burrowing underground to hide from the larger and more predatory therapsids that dominated the world. As food supply went away and all those bigger animals died off, these small cynodonts were among the few surviving therapsids to emerge on the other side of the Permian-Triassic extinction.
尽管兽孔目动物的血统被小型犬齿兽类动物勉强保留了下来，但它们所处的世界却截然不同。在这次灭绝事件的另一边，70% 的陆地生物灭绝，爬行动物数量众多、种类繁多、体型庞大。大型兽孔目动物的灭绝将动物王国交给了它们鳞片状的爬行动物表亲。从这次灭绝事件结束到接下来的一亿五千万年，爬行动物将统治世界。
Although the therapsid lineage was just barely preserved by the small cynodont, the world they found themselves in was different. On the other side of this extinction event, with 70 percent of land life extinguished, reptiles emerged numerous, diverse, and big. The eradication of the large therapsids handed the animal kingdom to their scaly reptilian cousins. From the end of this extinction event and for the next one hundred fifty million years, reptiles would rule.
二叠纪的小型蜥蜴进化成为 20 英尺长的掠食性主龙，拥有巨大的牙齿和爪子，类似于小型的暴龙。也是在这一时期，脊椎动物开始飞上天空——翼龙是一种会飞的主龙，是第一个长出翅膀并在空中捕猎的动物。
Small lizards of the Permian evolved into twenty-foot-long predatory archosaurs with massive teeth and claws, resembling a smaller Tyrannosaurus. It was also during this period that vertebrates took to the skies—the pterosaur, a flying archosaur, was the first to grow wings and hunt from above.
为了在掠食性恐龙、翼龙和其他大型爬行动物横行的时代生存下来，犬齿兽的体型越来越小，直到体长不超过四英寸。犬齿兽具备温血和小型化的特征，白天躲在洞穴中，在寒冷的夜晚出来生存，因为那时主龙类相对来说视力较差，行动不便。它们在挖好的洞穴迷宫中或厚厚的树皮中安家。它们在黄昏时分悄悄地在森林地面和树枝间徘徊，寻找昆虫。它们成为了第一批哺乳动物。
In order to survive this ravenous era of predatory dinosaurs, pterosaurs, and other massive reptilian beasts, cynodonts got smaller and smaller until they were no more than four inches long. Equipped with warm-bloodedness and miniaturization, they survived by hiding in burrows during the day and emerging during the cold night when archosaurs were relatively blind and immobile. They made their homes in dug-out burrowed mazes or in the thick bark of trees. They hunted by quietly wandering the twilight forest floors and tree branches in search of insects. They became the first mammals.
图 10.2：从第一批脊椎动物到第一批哺乳动物的进化树。MYA = 百万年前。
Figure 10.2: The evolutionary tree from the first vertebrates to the first mammals. MYA = million years ago.
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
在恐龙统治的这一亿年中，这些小型哺乳动物在世界的角落里幸存下来，并在它们的技能库中又增加了一项生存技巧。它们进化出了一种新的认知能力，这是自寒武纪鱼类以来最大的神经创新。
At some point in this hundred-million-year reign of dinosaurs, as these small mammals survived tucked away in nooks and crannies of the world, they added one more survival trick to their repertoire. They evolved a new cognitive ability, the biggest neural innovation since the Cambrian fish.
通过模拟生存
Surviving by Simulating
这种早期的哺乳动物体长约四英寸，可能与今天的老鼠或松鼠相似，它们的力量不如恐龙或鸟类，也显然无法逃脱掠食者的袭击。它们的速度可能也比从天而降的祖龙或翼龙慢，或至少不快。但穴居和树栖生活方式确实让早期哺乳动物拥有一项独特的优势：它们可以先发制人。它们可以从地下洞穴或树枝后面环顾四周，发现远处的鸟儿和美味的昆虫，然后决定是否逃跑。这种先发制人的天赋在数亿年中一直未被开发。但最终一种神经创新出现了，并加以利用：通过一系列目前未知的事件，大脑皮层的一个区域转变为一个称为新皮质的新区域（“neo ” 代表“新”）。
This early four-inch-long mammal, likely resembling a mouse or squirrel of today, was not stronger than dinosaurs or birds and surely unable to fight its way out of a predatory assault. It was probably also slower, or at least no faster, than an archosaur or a pterosaur swooping down from the sky. But the burrowing and arboreal lifestyle did indeed give early mammals a singular advantage: they got to make the first move. From an underground burrow or from behind a tree branch, they got to look around, spot a faraway bird and a tasty insect, and decide whether to make a run for it. This gift of the first move was left unexploited for hundreds of millions of years. But eventually a neural innovation emerged to exploit it: a region of the cortex transformed, through a currently unknown series of events, into a new region called the neocortex (neo for “new”).
大脑皮层赋予了这只小老鼠超能力——能够在动作发生之前模拟动作。它可以望向从洞里延伸到美味昆虫的树枝网。它可以看到附近掠食鸟远处的眼睛。老鼠可以模拟走不同的路，模拟鸟在追赶它，昆虫跳走，然后选择最佳路径——在模拟中，它发现自己既活着又吃饱了。如果强化学习的早期脊椎动物具有边做边学的能力，那么早期哺乳动物就具有更令人印象深刻的先做后学的能力——通过想象来学习。
The neocortex gave this small mouse a superpower—the ability to simulate actions before they occurred. It could look out at a web of branches leading from its hole to a tasty insect. It could see the faraway eyes of a nearby predatory bird. The mouse could simulate going down different paths, simulate the bird chasing it and the insects hopping away, then pick the best path—the one that, in its simulation, it found itself both alive and well fed. If the reinforcement-learning early vertebrates got the power of learning by doing, then early mammals got the even more impressive power of learning before doing—of learning by imagining.
许多生物以前曾发现自己处于第一步——螃蟹躲在沙子下面，小鱼穿梭在珊瑚植物的叶子之间。那么为什么只有哺乳动物才出现模拟行为呢？
Many creatures had previously found themselves in positions of having the first move—crabs hide under sand and small fish weave between the leaves of coral plants. So then why was it only with mammals that simulating emerged?
据推测，模拟进化有两个要求。首先，你需要远距离视野——你需要能够看到很多周围的环境，才能使模拟路径卓有成效。在陆地上，即使在晚上，你的视野也能比现在远一百倍比在水下快得多。因此，鱼不会模拟和计划它们的动作，而是在有东西靠近时迅速做出反应（因此它们的中脑和后脑较大，而皮层相对较小）。
It has been speculated that there were two requirements for simulating to evolve. First, you need far-ranging vision—you need to be able to see a lot of your surroundings in order for simulating paths to be fruitful. On land, even at night, you can see up to one hundred times farther than you can underwater. Thus, fish opted not to simulate and plan their movements but instead to respond quickly whenever something came at them (hence their large midbrain and hindbrain, and comparatively smaller cortex).
第二个推测要求是温血。出于我们将在后面几章中看到的原因，模拟动作比皮层-基底神经节系统中的强化学习机制在计算上要昂贵得多，而且耗时也更长。神经元的电信号对温度高度敏感——在较低温度下，神经元的放电速度比在较高温度下慢得多。这意味着温血的一个副作用是哺乳动物的大脑可以比鱼或爬行动物的大脑运转得更快。这使得执行更复杂的计算成为可能。这就是为什么爬行动物尽管在陆地上具有远距离视力，却从未被赋予模拟的天赋。唯一显示出模拟动作能力的非哺乳动物是最先进化出温血动物的是鸟类。而鸟类是现存唯一一种独立进化出温血的非哺乳动物。
The second speculated requirement is warm-bloodedness. For reasons we will see in the next few chapters, simulating actions is astronomically more computationally expensive and time-consuming than the reinforcement-learning mechanisms in the cortex-basal-ganglia system. The electrical signaling of neurons is highly sensitive to temperature—at lower temperatures, neurons fire much more slowly than at warmer temperatures. This meant that a side effect of warm-bloodedness was that mammal brains could operate much faster than fish or reptile brains. This made it possible to perform substantially more complex computations. This is why reptiles, despite their long-range vision on land, were never endowed with the gift of simulating. The only nonmammals that have shown evidence of the ability to simulate actions and plan are birds. And birds are, conspicuously, the only nonmammal species alive today that independently evolved warm-bloodedness.
第一批哺乳动物的大脑内部
Inside the Brain of the First Mammals
在这几亿年的历史中，从鱼类出现在陆地上到恐龙的崛起，动物的形态、大小和器官都发生了巨大的变化。然而，有一样东西却出人意料地没有改变：大脑。
Throughout this several-hundred-million-year-long story, from the emergence of fish onto land to the rise of dinosaurs, there was an expansive diversification of animal shapes, sizes, and organs. And yet, there was one thing that was surprisingly unchanged: brains.
从早期脊椎动物到第一批四足动物，再到爬行动物和兽孔目动物，大脑在很大程度上停留在神经黑暗时代。进化满足于（或至少接受了）早期脊椎动物的强化学习大脑，并将重点转向调整其他生物结构——形成颌骨、盔甲、肺、更符合人体工程学的身体、温血、鳞片、毛皮和其他类似的形态变化。这就是为什么现代鱼类和现代爬行动物的大脑尽管经历了数亿年的进化分离，却仍然非常相似。*
From the early vertebrates to the first tetrapods to reptiles and therapsids, brains were largely stuck in a neural dark age. Evolution settled for, or at least was resigned to, the reinforcement-learning brain of the early vertebrates, and it shifted its focus toward tweaking other biological structures—creating jaws, armor, lungs, more ergonomic bodies, warm-bloodedness, scales, fur, and other such morphological modifications. This is why the brain of a modern fish and that of a modern reptile, despite hundreds of millions of years of evolutionary separation, are remarkably similar.*
只有在早期哺乳动物中，创新的火花才从永恒的神经停滞中迸发出来。鱼类的皮层在早期哺乳动物中分裂成四个独立的结构，其中三个实际上与之前出现的亚区相同，只有新皮层这一部分可以真正被视为新的。早期脊椎动物的腹侧皮层成为哺乳动物的联想杏仁核，包含类似的回路，并且大致具有相同的用途：学习识别各种模态的模式，尤其是那些可以预测效价结果的模式（例如，预测声音 A 会导致好事，声音 B 会导致坏事）。早期脊椎动物外侧皮层中的嗅觉模式检测器成为哺乳动物的嗅觉皮层，其工作方式相同 - 通过自联想网络检测气味模式。早期脊椎动物的内侧皮层是学习空间地图的地方，它成为哺乳动物的海马体，使用类似的电路执行类似的功能。但大脑皮层的第四区域经历了更有意义的变化——它转变为新大脑皮层，其中包含完全不同的回路。
It was only in early mammals that a spark of innovation emerged from the eternity of neural stagnation. The fish cortex split into four separate structures in early mammals, three of which were effectively the same as the subregions that had come before, only one of which, the neocortex, could truly be considered new. The ventral cortex of early vertebrates became the associative amygdala in mammals, containing similar circuitry and serving largely the same purpose: learning to recognize patterns across various modalities, especially those that were predictive of valence outcomes (e.g., predicting that sound A leads to good things and sound B leads to bad things). The smell-pattern detectors in the lateral cortex of early vertebrates became the olfactory cortex in mammals, working the same way—detecting smell patterns through auto-associative networks. The medial cortex of early vertebrates, where spatial maps were learned, became the hippocampus of mammals, performing a similar function using similar circuitry. But a fourth region of the cortex underwent a more meaningful change—it transformed into the neocortex, which contained completely different circuitry.
除了新皮层的出现外，早期哺乳动物的大脑与早期脊椎动物的大脑大致相同。基底神经节整合了来自嗅觉皮层、海马体、杏仁核以及现在的新皮层有关世界的输入，以学习采取最大化多巴胺释放的行动。下丘脑仍然触发直接的价态反应，并通过多巴胺等神经调节剂调节其他结构。中脑和后脑结构仍然实施反射性运动模式，尽管现在专门用于行走而不是游泳。
Other than the emergence of the neocortex, the brain of early mammals was largely the same as that of early vertebrates. The basal ganglia integrated input about the world from the olfactory cortex, hippocampus, amygdala, and now also the neocortex to learn to take actions that maximized dopamine release. The hypothalamus still triggered direct valence responses and modulated other structures through neuromodulators such as dopamine. Midbrain and hindbrain structures still implemented reflexive movement patterns, albeit now specialized for walking as opposed to swimming.
图 10.3：从早期脊椎动物到早期哺乳动物的转变过程中皮层如何变化
Figure 10.3: How the cortex changed in the transition from early vertebrates to early mammals
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
这种早期哺乳动物的大脑皮层很小，只占大脑的一小部分。大部分体积都给了嗅觉皮层（早期哺乳动物和许多现代哺乳动物一样，嗅觉非常灵敏）。但尽管早期哺乳动物的大脑皮层体积很小，但它仍然是人类智慧产生的核心。在人类大脑中，大脑皮层占大脑体积的 70%。在随后的突破中，这个原本很小的结构逐渐从一个巧妙的技巧扩展到了智慧的中心。
The neocortex of this early mammal was small and took up only a small fraction of the brain. Most volume was given to the olfactory cortex (early mammals, like many modern mammals, had an incredible sense of smell). But despite the small size of the neocortex in early mammals, it was still the kernel from which human intelligence would arise. In the human brain, the neocortex takes up 70 percent of brain volume. In the breakthroughs that followed, this originally small structure would progressively expand from a clever trick to the epicenter of intelligence.
11
11
生成模型和新皮层之谜
Generative Models and the Neocortical Mystery
当你观察人类大脑时，你看到的几乎所有东西都是新皮质。新皮质是一张大约 2 到 4 毫米厚的薄片。随着新皮质变大，这片薄片的表面积也扩大了。为了适合头骨，它被折叠起来，就像你把毛巾捆起来放进手提箱一样。如果你展开一张人类新皮质薄片，它的表面积将近 3 平方英尺——大约是一张小桌子的大小。
WHEN YOU LOOK at a human brain, almost everything you see is neocortex. The neocortex is a sheet about two to four millimeters thick. As the neocortex got bigger, the surface area of this sheet expanded. To fit in the skull, it became folded, the way you would bunch up a towel to fit it in a suitcase. If you unfolded a human neocortical sheet, it would be almost three square feet in surface area—about the size of a small desk.
早期的实验得出的结论是，大脑皮层不承担任何单一的功能，而是承担多种不同的功能。例如，大脑皮层的后部处理视觉输入，因此被称为视觉皮层。*如果切除视觉皮层，你就会失明。如果你记录视觉皮层中神经元的活动，它们会对特定位置的特定视觉特征做出反应，例如某些颜色或线条方向。如果你刺激视觉皮层内的神经元，人们会报告看到闪光。
Early experimentation led to the conclusion that the neocortex didn’t serve any one function and instead subserved a multitude of different functions. For example, the back of the neocortex processes visual input and hence is called the visual cortex.* If you removed your visual cortex, you would become blind. If you record the activity of neurons in the visual cortex, they respond to specific visual features at specific locations, such as certain colors or line orientations. If you stimulate neurons within the visual cortex, people will report seeing flashes of lights.
在附近的一个区域，即听觉皮层，听觉感知也会发生同样的情况。听觉皮层受损会削弱人感知和理解声音的能力。如果你记录下听觉皮层神经元的活动，你会发现它们对特定的声音的频率。如果你刺激听觉皮层内的某些神经元，人们就会报告听到噪音。
In a nearby region called the auditory cortex, the same thing occurs with auditory perception. Damage to one’s auditory cortex impairs one’s ability to perceive and understand sounds. If you record the activity of neurons in the auditory cortex, you’ll find they are responsive to specific frequencies of sound. If you stimulate certain neurons within the auditory cortex, people will report hearing noises.
图 11.1：人类大脑皮层
Figure 11.1: The human neocortex
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
大脑皮层中还有其他区域负责触觉、疼痛和味觉。大脑皮层中还有其他区域似乎负责更加不同的功能——有负责运动、语言和音乐的区域。
There are other neocortical regions for touch, pain, and taste. And there are other areas of the neocortex that seem to serve even more disparate functions—there are areas for movement, language, and music.
乍一看，这毫无道理。一个结构怎么能做这么多不同的事情？
At first glance, this makes no sense. How can one structure do so many different things?
蒙特卡斯尔的疯狂想法
Mountcastle’s Crazy Idea
20 世纪中叶，神经科学家 Vernon Mountcastle 开创了当时一种新的研究范式：记录清醒动物大脑皮层中单个神经元的活动。这种新方法为动物在日常生活中大脑内部的运作提供了一个新视角。他用电极记录猴子体感皮层（处理触觉输入的大脑皮层区域）中的神经元，以观察触觉类型刺激引起什么反应。
In the mid-twentieth century, the neuroscientist Vernon Mountcastle was pioneering what was, at the time, a new research paradigm: recording the activity of individual neurons in the neocortex of awake-behaving animals. This new approach offered a novel view of the inner workings of brains as animals went about their life. He used electrodes to record the neurons in the somatosensory cortices (the neocortical area that processes touch input) of monkeys to see what types of touch stimuli elicited what responses.
Mountcastle 的第一个观察结果是，新皮质层垂直柱（直径约 500 微米）内的神经元似乎都对感觉刺激做出类似反应，而水平方向较远的神经元则没有。例如，视觉皮层内的单个柱可能包含对视野中特定位置的特定方向的光条做出类似反应的神经元。然而，附近柱内的神经元只对特定方向的光条做出反应。不同方向或位置的光条会做出反应。这一发现已在多种模式下得到证实。在大鼠中，大脑皮层中的某些柱状结构只对特定单个胡须的触碰做出反应，而每个相邻的柱状结构又对完全不同的胡须做出反应。在听觉大脑皮层中，某些柱状结构对特定频率的声音有选择性。
One of the first observations Mountcastle made was that neurons within a vertical column (about five hundred microns in diameter) of the neocortical sheet seemed to all respond similarly to sensory stimuli, while neurons horizontally farther away did not. For example, an individual column within the visual cortex might contain neurons that all similarly responded to bars of light at specific orientations at a specific location in the visual field. However, neurons within nearby columns responded only to bars of light at different orientations or locations. This same finding has been confirmed within multiple modalities. In rats, there are columns of neocortex that respond only to the touch of a specific single whisker, with each nearby column responding to a completely different whisker. In the auditory neocortex, there are individual columns that are selective for specific frequencies of sound.
图 11.2：新皮质柱
Figure 11.2: The neocortical column
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
Mountcastle 的第二个观察结果是，柱内垂直连接很多，而柱间连接相对较少。
The second observation that Mountcastle made was that there were many connections vertically within a column and comparatively fewer connections between columns.
蒙特卡斯尔的第三个也是最后一个观察结果是，在显微镜下，大脑新皮层在各个地方看起来都大体相同。听觉新皮层、躯体感觉新皮层和视觉新皮层都包含相同类型的神经元，并且以相同的方式组织。哺乳动物物种之间也是如此——在显微镜下，老鼠、猴子和人类的大脑新皮层看起来都相对相同。
The third and final observation Mountcastle made was that under a microscope, the neocortex looked largely identical everywhere. The auditory neocortex, somatosensory neocortex, and visual neocortex all contain the same types of neurons organized in the same way. And this is true across species of mammals—the neocortex of a rat, a monkey, and a human all look relatively the same under a microscope.
这三个事实——垂直排列的活动、垂直排列的连接性，以及观察到的大脑皮层所有区域之间的相似性——导致蒙特卡斯尔得出了一个惊人的结论：新皮质是由一个重复的微电路组成的，他称之为新皮质柱。皮质片只是一堆紧密堆积在一起的新皮质柱。
These three facts—vertically aligned activity, vertically aligned connectivity, and observed similarity between all areas of neocortex—led Mountcastle to a remarkable conclusion: the neocortex was made up of a repeating and duplicated microcircuit, what he called the neocortical column. The cortical sheet was just a bunch of neocortical columns packed densely together.
这为一个结构如何能做这么多不同的事情这一问题提供了一个令人惊讶的答案。根据蒙特卡斯尔的说法，大脑皮层不会做不同的事情；每个大脑皮层柱都做完全相同的事情。大脑皮层各区域之间的唯一区别在于它们接收的输入以及它们将输出发送到何处；大脑皮层本身的实际计算是相同的。例如，视觉皮层和听觉皮层之间的唯一区别在于视觉皮层从视网膜获取输入，而听觉皮层从耳朵获取输入。
This provided a surprising answer to the question of how one structure can do so many different things. According to Mountcastle, the neocortex does not do different things; each neocortical column does exactly the same thing. The only difference between regions of neocortex is the input they receive and where they send their output; the actual computations of the neocortex itself are identical. The only difference between, for example, the visual cortex and the auditory cortex is that the visual cortex gets input from the retina, and the auditory cortex gets input from the ear.
2000 年，即蒙特卡斯尔首次发表他的理论几十年后，麻省理工学院的三位神经科学家进行了一次精彩的测试 Mountcastle 的假设。如果大脑新皮质在任何地方都一样，如果视觉皮质和听觉皮质都不是独一无二的，那么你会认为这些区域是可以互换的。在对幼年雪貂进行实验时，科学家们切断了耳朵的输入，并将视网膜的输入重新连接到听觉皮质而不是视觉皮质。如果 Mountcastle 错了，雪貂最终会失明或视力受损——听觉皮质无法正确处理眼睛的输入。如果大脑新皮质确实在任何地方都一样，那么接收视觉输入的听觉皮质应该与视觉皮质的工作方式相同。
In the year 2000, decades after Mountcastle first published his theory, three neuroscientists at MIT performed a brilliant test of Mountcastle’s hypothesis. If the neocortex is the same everywhere, if there is nothing uniquely visual about the visual cortex or auditory about the auditory cortex, then you would expect these areas to be interchangeable. Experimenting on young ferrets, the scientists cut off input from the ears and rewired input from the retina to the auditory cortex instead of the visual cortex. If Mountcastle was wrong, the ferrets would end up blind or visually impaired—input from the eye in the auditory cortex would not be processed correctly. If the neocortex was indeed the same everywhere, then the auditory cortex receiving visual input should work the same way as the visual cortex.
值得注意的是，雪貂的视力非常好。当研究人员记录大脑皮层中原本负责听觉但现在接收来自眼睛的输入的区域时，他们发现该区域对视觉刺激的反应与视觉皮层一样。听觉皮层和视觉皮层是可以互换的。
Remarkably, the ferrets could see just fine. And when researchers recorded the area of the neocortex that was typically auditory but was now receiving input from the eyes, they found the area responded to visual stimuli just as the visual cortex would. The auditory and visual cortices are interchangeable.
对先天性失明患者的研究进一步证实了这一点，他们的视网膜从未向大脑发送过任何信号。在这些患者中，视觉皮层从未接受过来自眼睛的输入。然而，如果你记录先天性失明的人视觉皮层神经元的活动，你会发现视觉皮层并没有变得毫无功能区域。相反，它会对大量其他感官输入做出反应，例如声音和触觉。这进一步证实了盲人确实拥有优越的听力——视觉皮层被重新用于帮助听觉。同样，新皮质的各个区域似乎是可以互换的。
This was further reinforced by studies of congenitally blind patients whose retinas had never sent any signals to their brains. In these patients, the visual cortex never received input from the eyes. However, if you record the activity of neurons in the visual cortex of congenitally blind humans, you find that the visual cortex has not been rendered a functionally useless region. Instead, it becomes responsive to a multitude of other sensory input, such as sounds and touch. This puts meat on the bone of the idea that people who are blind do, in fact, have superior hearing—the visual cortex becomes repurposed to aid in audition. Again, areas of neocortex seem interchangeable.
以中风患者为例。当患者的大脑皮层特定区域受损时，他们会立即失去该区域的功能。如果运动皮层受损，患者可能会瘫痪。如果视觉皮层受损，患者会部分失明。但随着时间的推移，功能可以恢复。这通常不是大脑皮层受损区域恢复的结果；通常，大脑皮层的那个区域将永远处于死亡状态。相反，大脑皮层的附近区域会被重新利用，以履行现已受损的大脑皮层区域的功能。这也表明大脑皮层的各区域是可以互换的。
Consider stroke patients. When patients have damage to a specific area of neocortex, they immediately lose the function in that area. If the motor cortex is damaged, patients can become paralyzed. If the visual cortex is damaged, patients become partially blind. But over time, function can return. This is usually not the consequence of the damaged area of neocortex recovering; typically, that area of neocortex remains dead forever. Instead, nearby areas of neocortex become repurposed to fulfill the functions of the now-damaged area of neocortex. This too suggests that areas of neocortex are interchangeable.
对于人工智能界的人们来说，芒特卡斯尔的假说是一份无与伦比的科学礼物。人类大脑皮层由超过一百亿个神经元和数万亿个连接组成；试图破译如此庞大的神经元毛团所执行的算法和计算是一项毫无希望的努力。如此无望以至于许多神经科学家认为试图破译大脑皮层的工作方式是一项徒劳无功的努力，注定会失败。但芒特卡斯尔的理论提供了一个更有希望的研究议程——我们不必试图了解整个人类大脑皮层，也许我们只需要了解重复一百万次左右的微电路的功能。我们不必了解整个大脑皮层中的数万亿个连接，也许我们只需要了解大脑皮层柱内的一百万个左右的连接。此外，如果芒特卡斯尔的理论是正确的，它表明大脑皮层柱实现了一些非常通用的算法，可以应用于各种功能，例如运动、语言和跨每种感觉模式的感知。
To those in the AI community, Mountcastle’s hypothesis is a scientific gift like no other. The human neocortex is made up of over ten billion neurons and trillions of connections; it is a hopeless endeavor to try and decode the algorithms and computations performed by such an astronomically massive hairball of neurons. So hopeless that many neuroscientists believe that attempting to decode how the neocortex works is a fruitless endeavor, doomed to fail. But Mountcastle’s theory offers a more hopeful research agenda—instead of trying to understand the entire human neocortex, perhaps we only have to understand the function of the microcircuit that is repeated a million or so times. Instead of understanding the trillions of connections in the entire neocortex, perhaps we only have to understand the million or so connections within the neocortical column. Further, if Mountcastle’s theory is correct, it suggests that the neocortical column implements some algorithm that is so general and universal that it can be applied to extremely diverse functions such as movement, language, and perception across every sensory modality.
在显微镜下可以看到这种微电路的基本结构。新皮质包含六层神经元（与早期脊椎动物的三层皮质不同）。这六层神经元以复杂但一致的方式连接。有一种特定类型的第五层中的神经元总是投射到基底神经节、丘脑和运动区。在第四层中，有一些神经元总是直接从丘脑获得输入。在第六层中，有一些神经元总是投射到丘脑。它不仅仅是一堆随机连接的神经元；微电路以特定方式预先连接以执行某些特定计算。
The basics of this microcircuit can be seen under a microscope. The neocortex contains six layers of neurons (unlike the three-layered cortex seen in earlier vertebrates). These six layers of neurons are connected in a complicated but beautifully consistent way. There is a specific type of neuron in layer five that always projects to the basal ganglia, the thalamus, and the motor areas. In layer four, there are neurons that always get input directly from the thalamus. In layer six, there are neurons that always project to the thalamus. It is not just a soup of randomly connected neurons; the microcircuit is prewired in a specific way to perform some specific computation.
图 11.3：新皮质柱的微电路
Figure 11.3: The microcircuitry of the neocortical column
原创艺术由 Mesa Schumacher 创作
Original art by Mesa Schumacher
当然，问题是：计算是什么？
The question is, of course: What is the computation?
感知的特殊性质
Peculiar Properties of Perception
十九世纪，人类感知的科学研究开始全面展开。世界各地的科学家开始探索人类的思维。视觉是如何运作的？听觉是如何运作的？
In the nineteenth century, the scientific study of human perception began in full force. Scientists around the world started probing the mind. How does vision work? How does audition work?
对感知的探究始于幻觉的使用；通过操纵人们的视觉感知，科学家发现了感知的三个奇特属性。而且由于大部分感知（至少对人类而言）发生在大脑皮层中，这些感知属性可以让我们了解大脑皮层是如何运作的。
The inquiry into perception began with the use of illusions; by manipulating people’s visual perceptions, scientists uncovered three peculiar properties of perception. And because much of perception, in humans at least, occurs in the neocortex, these properties of perception teach us about how the neocortex works.
属性 #1：填充
Property #1: Filling In
这些 19 世纪的科学家首先意识到，人类大脑会自动且无意识地填补缺失的内容。请看图 11.4中的图像。你立即看到了“editor”这个词。但这并不是你的眼睛实际看到的内容——字母的大部分线条都缺失了。在其他图像中，你的大脑也看到了一些不存在的东西：一个三角形、一个球体和一个被某种东西缠绕着的条形物。
The first thing that became clear to these nineteenth-century scientists was that the human mind automatically and unconsciously fills in missing things. Consider the images in figure 11.4. You immediately perceive the word editor. But this is not what your eye is actually seeing—most of the lines of the letters are missing. In the other images too, your mind perceives something that is not there: a triangle, a sphere, and a bar with something wrapped around it.
这种填充并不是视觉所独有的特性；它存在于我们大多数感官模式中。这就是为什么你仍然可以通过混乱的电话连接理解别人在说什么，以及为什么即使闭上眼睛也能通过触觉识别物体。
This filling in is not a property unique to vision; it is seen across most of our sensory modalities. This is how you can still understand what someone is saying through a garbled phone connection and how you can identify an object through touch even with your eyes closed.
Figure 11.4: Filling-in property of perception
“编辑”摘自 Jastrow，1899 年。其他摘自 Lehar，2003 年。
“Editor” from Jastrow, 1899. Others from Lehar, 2003.
属性 #2：一次一个
Property #2: One at a Time
如果你的大脑根据感官证据来填补它认为存在的东西，那么如果有多种方式来填补你看到的东西会发生什么？图 11.5中的所有三个图像都是 19 世纪为探究这个问题而设计的视觉错觉的例子。这些图像中的每一个都可以用两种不同的方式来解释。在图 11.5的左侧，你可以将它看作楼梯，但你也可以将它看作楼梯下方的突出部分。 *在图 11.5的中间，立方体可以是右下角正方形是正面，或者左上角的方块是正面。在图 11.5的右侧，图片可能是一只兔子或一只鸭子。
If your mind fills in what it thinks is there based on sensory evidence, what happens if there are multiple ways to fill in what you see? All three of the images in figure 11.5 are examples of visual illusions devised in the 1800s to probe this question. Each of these images can be interpreted in two different ways. On the left side of figure 11.5, you can see it as a staircase, but you can also see it as a protrusion from under a staircase.* In the middle of figure 11.5, the cube could be one where the bottom right square is the front or where the top left square is the front. On the right of figure 11.5, the picture could be a rabbit or a duck.
Figure 11.5: One-at-a-time property of perception
施罗德 (Schroeder) 的楼梯，1858 年。“内克尔立方体”来自内克尔 (Necker)，1832 年。鸭子或兔子来自贾斯特罗 (Jastrow)，1899 年。
Staircase from Schroeder, 1858. “Necker cube” from Necker, 1832. Duck or rabbit from from Jastrow, 1899.
所有这些模棱两可的图片的有趣之处在于，你的大脑一次只能看到一种解释。你无法同时看到一只鸭子和一只兔子，即使感官证据同样暗示着两者。出于某种原因，大脑的感知机制要求它只能选择其中一种。
What is interesting about all these ambiguous pictures is that your brain can see only one interpretation at a time. You cannot see a duck and a rabbit simultaneously, even though the sensory evidence is equally suggestive of both. The mechanisms of perception in the brain, for some reason, require it to pick only one.
这也适用于听觉。考虑一下“鸡尾酒会效应”。如果你参加一个吵闹的鸡尾酒会，你可以收听你正在交谈的人的谈话或附近一群人的谈话。但你不能同时听到两个对话。无论你收听哪一场对话，进入你耳朵的听觉输入都是相同的；唯一的区别是你的大脑从该输入中推断出什么。你一次只能听到一场对话。
This also applies to audition. Consider the “cocktail-party effect.” If you are at a noisy cocktail party, you can tune in to the conversation of the person you are speaking to or the conversation of a nearby group. But you cannot listen to both conversations at the same time. No matter which conversation you tune in to, the auditory input into your ear is identical; the only difference is what your brain infers from that input. You can perceive only a single conversation at a time.
属性 #3：无法忽视
Property #3: Can’t Unsee
当感官证据模糊不清时会发生什么——当它无法被解释为任何有意义的东西时会发生什么？考虑 11.6 中的图像。如果你以前没见过这些，它们看起来什么都不是——只是斑点。如果我给你一个对这些斑点的合理解释，你对它们的感知就会突然改变。
What happens when sensory evidence is vague—when it isn’t clear that it can be interpreted as anything meaningful at all? Consider the image in 11.6. If you haven’t seen these before, they will look like nothing—just blobs. If I give you a reasonable interpretation of these blobs, all of a sudden, your perception of them will change.
图 11.6可以解释为一只青蛙（如果您不明白，请参见此处）。一旦您的大脑感知到这种解释，您将永远无法忽视它。这可以称为感知的“无法忽视”属性。您的大脑喜欢有一种解释来解释感官输入。一旦我给您一个很好的解释，您的大脑就会坚持下去。您现在感知到的是一只青蛙。
Figure 11.6 can be interpreted as a frog (see here if you don’t see this). Once your mind perceives this interpretation, you will never be able to unsee it. This is what might be called the can’t-unsee property of perception. Your mind likes to have an interpretation that explains sensory input. Once I give you a good explanation, your mind sticks to it. You now perceive a frog.
Figure 11.6: The can’t-unsee property of perception
图片来自 Fahle 等人，2002 年。经麻省理工学院出版社许可使用。
Image from Fahle et al., 2002. Used with permission by The MIT Press.
十九世纪，德国物理学家兼医生赫尔曼·冯·亥姆霍兹提出了一种新理论来解释感知的这些特性。他认为，人不会感知到自己所经历的事情，而是感知到大脑认为存在的东西——亥姆霍兹将这一过程称为推理。换句话说：你感知的不是你实际看到的东西，而是你从所见事物中推断出的模拟现实。
In the nineteenth century, a German physicist and physician named Hermann von Helmholtz proposed a novel theory to explain these properties of perception. He suggested that a person doesn’t perceive what is experienced; instead, he or she perceives what the brain thinks is there—a process Helmholtz called inference. Put another way: you don’t perceive what you actually see, you perceive a simulated reality that you have inferred from what you see.
这个想法解释了感知的所有这三种奇特属性。你的大脑会填补物体缺失的部分，因为它试图解读你的视觉所暗示的真相（“那里真的有一个球体吗？”）。你一次只能看到一件事，因为你的大脑必须选择一个现实来模拟——事实上，动物不可能既是兔子又是鸭子。一旦你发现一幅图像最好解释为青蛙，你的大脑在观察它时就会保持这个现实。
This idea explains all three of these peculiar properties of perception. Your brain fills in missing parts of objects because it is trying to decipher the truth that your vision is suggesting (“Is there actually a sphere there?”). You can see only one thing at a time because your brain must pick a single reality to simulate—in reality, the animal can’t be both a rabbit and a duck. And once you see that an image is best explained as a frog, your brain maintains this reality when observing it.
虽然许多心理学家原则上同意，与亥姆霍兹的理论不同，又过了一个世纪才有人提出亥姆霍兹的推理感知实际上如何起作用。
While many psychologists came to agree, in principle, with Helmholtz’s theory, it would take another century before anyone proposed how Helmholtz’s perception by inference might actually work.
生成模型：通过模拟进行识别
Generative Models: Recognizing by Simulating
20 世纪 90 年代，Geoffrey Hinton 和他的一些学生（包括帮助发现多巴胺反应是时间差分学习信号 (temporal difference learning signals) 的研究者们立志要建立一个按照亥姆霍兹所建议的方式进行学习的人工智能系统。
In the 1990s, Geoffrey Hinton and some of his students (including the same Peter Dayan that had helped discover that dopamine responses are temporal difference learning signals) set their sights on building an AI system that learned in the way that Helmholtz suggested.
我们在第 7 章中回顾了大多数现代神经网络是如何在监督下进行训练的：将一张图片（例如，一张狗的图片）连同正确答案（例如，“这是一只狗”）一起提供给网络，网络中的连接被推向正确的方向，以使其给出正确的答案。大脑不太可能以这种方式使用监督来识别物体和模式。大脑必须以某种方式识别世界的各个方面，而无需被告知正确答案；它们必须进行无监督学习。
We reviewed in chapter 7 how most modern neural networks are trained with supervision: a picture is given to a network (e.g., a picture of a dog) along with the correct answer (e.g., “This is a dog”), and the connections in the network are nudged in the right direction to get it to give the right answer. It is unlikely the brain recognizes objects and patterns using supervision in this way. Brains must somehow recognize aspects of the world without being told the right answer; they must engage in unsupervised learning.
一类无监督学习方法是自联想网络，就像我们推测在早期脊椎动物的皮层中出现的网络一样。基于输入模式的相关性，这些网络将常见的输入模式聚类到神经元集合中，从而提供了一种方法，可以将重叠模式识别为不同的模式，并可以完成嘈杂和受阻的模式。
One class of unsupervised-learning methods are auto-associative networks, like those we speculated emerged in the cortex of early vertebrates. Based on correlations in input patterns, these networks cluster common patterns of input into ensembles of neurons, offering a way in which overlapping patterns can be recognized as distinct, and noisy and obstructed patterns can be completed.
但亥姆霍兹认为，人类感知的作用远不止这些。他认为，人类感知并非只是根据相关性对输入模式进行聚类，而是可以优化内部模拟现实预测当前外部感官输入的准确性。
But Helmholtz suggested that human perception was doing something more than this. He suggested that instead of simply clustering incoming input patterns based on their correlations, human perception might optimize for the accuracy with which the inner simulated reality predicts the current external sensory input.
1995 年，Hinton 和 Dayan 提出了一个概念证明亥姆霍兹通过推理感知的思想；他们将其命名为亥姆霍兹机。亥姆霍兹机在原理上与其他神经网络类似；它接收从一端流向另一端的输入。但与其他神经网络不同的是，它还具有反向连接，即从一端流向另一端。
In 1995, Hinton and Dayan came up with a proof of concept for Helmholtz’s idea of perception by inference; they named it the Helmholtz machine. The Helmholtz machine was, in principle, similar to other neural networks; it received inputs that flowed from one end to the other. But unlike other neural networks, it also had backward connections that flowed the opposite way—from the end to the beginning.
Hinton 使用 0 到 9 之间的手写数字图像测试了这个网络。可以在网络底部给出手写数字的图片（每个像素一个神经元），它将向上流动并激活顶部的一组随机神经元。然后，这些顶部激活的神经元可以流回并激活底部的一组神经元以产生自己的图片。学习旨在使网络稳定到这样一种状态：流入网络的输入在流回时被准确地重新创建。
Hinton tested this network with images of handwritten numbers between 0 and 9. A picture of a handwritten number can be given at the bottom of the network (one neuron for each pixel) and will flow upward and activate a random set of neurons at the top. These activated neurons at the top can then flow back down and activate a set of neurons at the bottom to produce a picture of its own. Learning was designed to get the network to stabilize to a state where input that flows up the network is accurately re-created when it flows back down.
一开始，流入的图像神经元的值和流出的结果之间会有很大差异。Hinton 设计了这个网络以两种独立的模式进行学习：识别模式和生成模式。在识别模式下，信息在网络中向上流动（从一张数字 7 的输入图片开始，到顶部的一些神经元），后向权重被微调，以使网络顶部激活的神经元更好地重现输入的传感数据（模拟出一个很好的数字 7）。相反，在生成模式下，信息在网络中向下流动（从生成一个想象的数字 7 的图片的目标开始），前向权重被微调，以使网络底部激活的神经元在顶部得到正确识别（“我将我刚刚生成的数字识别为 7”）。
At first, there will be big discrepancies between the values in neurons from the image flowing in and the result flowing out. Hinton designed this network to learn with two separate modes: recognition mode and generative mode. When in recognition mode, information flows up the network (starting from an input picture of a 7 to some neurons at the top), and the backward weights are nudged to make the neurons activated at the top of the network better reproduce the input sensory data (make a good simulated 7). In contrast, when in generative mode, information flows down the network (starting from the goal to produce an imagined picture of a 7), and the forward weights are nudged so that the neurons activated at the bottom of the network are correctly recognized at the top (“I recognize what I just made as a 7”).
没有人告诉这个网络正确的答案；它从未被告知什么属性构成了 2，甚至不知道哪些图片是 2、7 或任何其他数字。网络唯一需要学习的数据是数字图片。当然，问题是，这行得通吗？在识别和生成之间来回切换是否能让网络既识别手写数字，又生成自己独特的手写数字图片，而无需被告知正确答案？
Nowhere was this network told the right answer; it was never told what properties make up a 2 or even which pictures were 2s or 7s or any other number. The only data the network had to learn from was pictures of numbers. The question was, of course, would this work? Would this toggling back and forth between recognition and generation enable the network to both recognize handwritten numbers and generate its own unique pictures of handwritten numbers without ever having been told the right answer?
令人惊奇的是，它确实做到了；它它可以自行学习。当这两个过程来回切换时，网络会神奇地稳定下来。当你给它一张数字 7 的图片时，它在很大程度上能够创建一个类似的向下移动时会生成 7 的图像。如果您给它一个 8 的图像，它将能够重新生成 8 的输入图像。
Amazingly, it did; it learned on its own. When these two processes toggle back and forth, the network magically stabilizes. When you give it a picture of the number 7, it will be able to, for the most part, create a similar image of a 7 on the way down. If instead you give it an image of an 8, it will be able to regenerate an input image of an 8.
图 11.7：亥姆霍兹机
Figure 11.7: The Helmholtz Machine
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
这似乎并不特别令人印象深刻。你给一个网络一张数字的图片，它吐出了一张相同数字的图片——这有什么大不了的？这个网络有三个开创性的特征。首先，这个网络的顶部现在可以在没有任何监督的情况下可靠地“识别”不完美的手写字母。其次，它的泛化能力令人印象深刻；它可以分辨出两张不同手写的 7 图片都是 7——它们将激活网络顶部的一组相似的神经元网络。第三，也是最重要的一点，这个网络现在可以生成手写数字的新图片。通过操纵这个网络顶部的神经元，你可以创建大量的手写 7 或手写 4 或它已经学会的任何数字。这个网络已经学会通过生成自己的数据来识别。
This might not seem particularly impressive. You gave a network a picture of a number, and it spit out a picture of that same number—what’s the big deal? There are three attributes of this network that are groundbreaking. First, the top of this network now reliably “recognizes” imperfectly handwritten letters without any supervision. Second, it generalizes impressively well; it can tell that two differently handwritten pictures of 7s are both a 7—they will activate a similar set of neurons at the top of the network. And third, and most important, this network can now generate novel pictures of handwritten numbers. By manipulating neurons at the top of this network, you can create lots of handwritten 7s or handwritten 4s or any number it has learned. This network has learned to recognize by generating its own data.
图片来自 Hinton 等人，1995 年。已获许可使用。
Image from Hinton et al., 1995. Used with permission.
亥姆霍兹机器是生成模型这一更广泛模型类别的早期概念证明。大多数现代生成模型都比亥姆霍兹机器复杂，但它们都具有一个基本属性，即它们通过生成自己的数据并将生成的数据与实际数据进行比较来学习识别世界上的事物。
The Helmholtz machine was an early proof of concept of a much broader class of models called generative models. Most modern generative models are more complicated than the Helmholtz machine, but they share the essential property that they learn to recognize things in the world by generating their own data and comparing the generated data to the actual data.
如果你对小像素手写数字的生成不感兴趣，请考虑一下这些生成模型自 1995 年以来取得了多大的进步。在这本书即将付印时，有一个活跃的网站叫做thispersondoesnotexist.com。每次刷新页面，你都会看到不同的人的照片。现实更令人震惊：每次重新加载页面时，生成模型都会创建一个全新的、从未见过的化妆脸。你看到的脸并不存在。
If you aren’t impressed with the generation of small pixelated handwritten numbers, consider how far these generative models have come since 1995. As this book is going to print, there is an active website called thispersondoesnotexist.com. Every time you refresh the page, you will see a picture of a different person. The reality is more shocking: every time you reload the page, a generative model creates a completely new, never before seen, made-up face. The faces you see do not exist.
这些生成模型的神奇之处在于，它们学会了捕捉输入的基本特征，而无需任何监督。生成逼真的新颖面孔的能力需要模型理解构成面孔的本质以及面孔可以变化的多种方式。就像激活亥姆霍兹机器顶部的各种神经元可以生成不同手写数字的图像一样，如果你激活这个面孔生成模型顶部的各种神经元，你就可以控制它生成的面孔类型。如果你改变一组神经元的值，网络就会吐出同一张经过旋转的面孔。如果你改变另一组神经元的值，它就会增加胡须、改变年龄或改变头发的颜色（见图11.10）。
What is so amazing about these generative models is that they learn to capture the essential features of the input they are given without any supervision. The ability to generate realistic novel faces requires the model to understand the essence of what constitutes a face and the many ways it can vary. Just as activating various neurons at the top of the Helmholtz machine can generate images of different handwritten numbers, if you activate various neurons at the top of this generative model of faces, you can control what types of faces it generates. If you change the value of one set of neurons, the network will spit out the same face but rotated. If you change the value of a different set of neurons, it will add a beard or change the age or alter the color of the hair (see figure 11.10).
图 11.9：来自 thispersondoesnotexist.com 的 StyleGAN2
Figure 11.9: StyleGAN2 from thispersondoesnotexist.com
图片来自thispersondoesnotexist.com
Pictures from thispersondoesnotexist.com
虽然 21 世纪初发生的大多数 AI 进步都涉及监督学习模型的应用，但最近的许多进步都是生成模型的应用。Deepfakes、AI 生成的艺术和 GPT-3 等语言模型都是生成模型发挥作用的例子。
While most AI advancements that occurred in the early 2000s involved applications of supervised-learning models, many of the recent advancements have been applications of generative models. Deepfakes, AI-generated art, and language models like GPT-3 are all examples of generative models at work.
亥姆霍兹认为，人类感知的大部分过程都是推理过程，即使用生成模型将内部模拟的世界与呈现的感官证据相匹配的过程。现代生成模型的成功为他的想法提供了依据；这些模型表明至少在理论上，这样的方法可以奏效。事实证明，有大量证据表明新皮层微电路正在实现这样的生成模型。
Helmholtz suggested that much of human perception is a process of inference—a process of using a generative model to match an inner simulation of the world to the sensory evidence presented. The success of modern generative models gives weight to his idea; these models reveal that something like this can work, at least in principle. It turns out that there is, in fact, an abundance of evidence that the neocortical microcircuit is implementing such a generative model.
Figure 11.10: Changing images by changing latent representations in generative models
图片来自 He et al.，2019。已获许可使用。
Figure from He et al., 2019. Used with permission.
证据是，一次填充一个，无法忽视的视觉错觉；证据是，大脑皮层本身的接线方式已被证明具有许多与大脑皮层一致的特性。生成模型；证据是感知和想象之间惊人的对称性——不可分割的不可分割性——在生成模型和大脑皮层中都存在。事实上，大脑皮层作为生成模型，不仅能解释视觉错觉——它还能解释人类为何会出现幻觉，为何会做梦和睡觉，甚至想象力本身的内部运作。
Evidence for this is seen in filling-in, one-at-a-time, can’t-unsee visual illusions; evidence is seen in the wiring of the neocortex itself, which has been shown to have many properties consistent with a generative model; and evidence is seen in the surprising symmetry—the ironclad inseparability—between perception and imagination that is found in both generative models and the neocortex. Indeed, the neocortex as a generative model explains more than just visual illusions—it also explains why humans succumb to hallucinations, why we dream and sleep, and even the inner workings of imagination itself.
幻觉、做梦和想象：大脑皮层作为生成模型
Hallucinating, Dreaming, and Imagining: The Neocortex as a Generative Model
无论是由于视神经损伤还是视网膜损伤，眼睛停止向大脑皮层发送信号的人，通常会患上一种叫做查尔斯·邦纳综合征的疾病。你可能会认为，当一个人的眼睛与大脑断开连接时，他们就再也看不见东西了。但事实恰恰相反——在失明后的几个月里，人们开始看到很多东西。他们开始产生幻觉。这种现象符合生成模型：切断大脑皮层的感觉输入会使其变得不稳定。它会陷入一个漂移的生成过程，在这个过程中，视觉场景的模拟不受实际感觉输入的限制——因此你产生了幻觉。
People whose eyes stop sending signals to their neocortex, whether due to optic-nerve damage or retinal damage, often get something called Charles Bonnet syndrome. You would think that when someone’s eyes are disconnected from their brain, they would no longer see. But the opposite happens—for several months after going blind, people start seeing a lot. They begin to hallucinate. This phenomenon is consistent with a generative model: cutting off sensory input to the neocortex makes it unstable. It gets stuck in a drifting generative process in which visual scenes are simulated without being constrained to actual sensory input—thus you hallucinate.
一些神经科学家认为，即使感知功能正常，它也是一种“受限幻觉”。如果没有感官输入，这种幻觉就会变得不受约束。在我们的亥姆霍兹机器示例中，这就像随机激活网络顶部的神经元并产生数字的幻觉图像，而从未将这些幻觉建立在真实的感官输入上。
Some neuroscientists refer to perception, even when it is functioning properly, as a “constrained hallucination.” Without sensory input, this hallucination becomes unconstrained. In our example of the Helmholtz machine, this is like randomly activating neurons at the top of the network and producing hallucinated images of numbers without ever grounding these hallucinations in real sensory input.
当然，这种将感知视为受限幻觉的想法正是亥姆霍兹所说的推理，也正是生成模型所做的。我们将内心对现实的幻觉与感官我们看到的数据。当视觉数据显示图片中有一个三角形时（即使实际上没有三角形），我们就会幻觉中出现一个三角形，因此产生填充效果。
This idea of perception as a constrained hallucination is, of course, exactly what Helmholtz meant by inference and exactly what a generative model is doing. We match our inner hallucination of reality to the sensory data we are seeing. When the visual data suggests there is a triangle in a picture (even if there is not actually a triangle there), we hallucinate a triangle, hence the filling-in effect.
生成模型或许还能解释我们为什么会做梦以及为什么需要睡眠。大多数动物都会睡觉，睡眠有很多好处，比如节省能源；但只有哺乳动物和鸟类表现出明确的做梦证据，这是通过睡眠不足会导致幻觉和感知紊乱。只有哺乳动物和鸟类在睡眠不足时才会出现幻觉和感知紊乱。事实上，鸟类似乎已经独立进化出了类似新皮层的结构。
Generative models may also explain why we dream and why we need sleep. Most animals sleep, and it has numerous benefits, such as saving energy; but only mammals and birds show unequivocal evidence of dreaming as measured by the presence of REM sleep. And it is only mammals and birds who exhibit hallucinations and disordered perception if deprived of sleep. Indeed, birds seem to have independently evolved their own neocortex-like structure.
新皮质（大概也是鸟类的对应部位）总是处于识别和生成之间的不稳定平衡，而在我们清醒的生活中，人类花费不平衡的时间用于识别，而相对较少的时间用于生成。也许梦就是对这种平衡的一种平衡，一种通过强迫生成的过程。如果我们被剥夺了睡眠，这种过多识别和生成不足的不平衡最终会变得非常严重，以至于大脑皮层中的生成模型变得不稳定。因此，哺乳动物开始产生幻觉，识别变得扭曲，生成和识别之间的区别变得模糊。恰如其分的是，Hinton 甚至将训练他的亥姆霍兹机器的学习算法称为“清醒-睡眠算法”。识别步骤是模型“清醒”时的步骤；生成步骤是模型“睡眠”时的步骤。
The neocortex (and presumably the bird equivalent) is always in an unstable balance between recognition and generation, and during our waking life, humans spend an unbalanced amount of time recognizing and comparatively less time generating. Perhaps dreams are a counterbalance to this, a way to stabilize the generative model through a process of forced generation. If we are deprived of sleep, this imbalance of too much recognition and not enough generation eventually becomes so severe that the generative model in the neocortex becomes unstable. Hence, mammals start hallucinating, recognition becomes distorted, and the difference between generation and recognition gets blurred. Fittingly, Hinton even called the learning algorithm to train his Helmholtz machine a “wake-sleep algorithm.” Recognition step was when the model was “awake”; the generation step was when the model was “asleep.”
哺乳动物想象力的许多特征与我们对生成模型的期望一致。人类很容易，甚至很自然地想象他们目前没有经历的事情。你可以想象你昨晚吃的晚餐，或者想象你今天晚些时候要做的事情。当你想象某事时，你在做什么？这只是你的大脑皮层处于生成模式。你正在大脑皮层中调用模拟现实。
Many features of imagination in mammals are consistent with what we would expect from a generative model. It is easy, even natural, for humans to imagine things that they are not currently experiencing. You can imagine the dinner you ate last night or imagine what you will be doing later today. What are you doing when you are imagining something? This is just your neocortex in generation mode. You are invoking a simulated reality in your neocortex.
想象力最明显的特征是，你无法同时想象事物和识别事物。你无法同时读书和想象自己吃早餐——想象的过程本质上与体验实际感官数据的过程相矛盾。事实上，你可以通过以下方式判断某人是否在想象某事看着那个人的瞳孔——当人们想象事物时，他们的瞳孔会随着大脑停止而扩大处理实际的视觉数据。人们会变成伪盲人。与生成模型一样，生成和识别不能同时进行。
The most obvious feature of imagination is that you cannot imagine things and recognize things simultaneously. You cannot read a book and imagine yourself having breakfast at the same time—the process of imagining is inherently at odds with the process of experiencing actual sensory data. In fact, you can tell when someone is imagining something by looking at that person’s pupils—when people are imagining things, their pupils dilate as their brains stop processing actual visual data. People become pseudo-blind. As in a generative model, generation and recognition cannot be performed simultaneously.
此外，如果你记录在识别过程中变得活跃的新皮层神经元（例如，对面部或房屋做出反应的神经元），那么当你只是想象同样的事情。当你想象移动身体的某些部位时，同样的区域会被激活，就像你真的移动身体部位。当你想象某些形状时，视觉皮层激活的区域与你看到这些形状时激活的区域相同。事实上，这是如此一致，以至于神经科学家可以通过记录他们的新皮层活动来解码人们的想象（作为做梦和想象是同一个一般过程的证据，科学家也可以通过以下方式准确解码人们的梦境记录他们的大脑）。大脑皮层受损会损害某些传感数据（例如无法识别视野左侧的物体），同样，那些仅仅想象相同传感数据特征的人也会受到影响（他们甚至很难想象左视野）。
Further, if you record neocortical neurons that become active during recognition (say, neurons that respond to faces or houses), those exact same neurons become active when you simply imagine the same thing. When you imagine moving certain body parts, the same area activates as if you were actually moving the body parts. When you imagine certain shapes, the same areas of visual cortex become activated as when you see those shapes. In fact, this is so consistent that neuroscientists can decode what people are imagining simply by recording their neocortical activity (and as evidence that dreaming and imagination are the same general process, scientists can also accurately decode people’s dreams by recording their brains). People with neocortical damage that impairs certain sensor data (such as being unable to recognize objects on the left side of the visual field) become equally impaired at simply imagining features of that same sensory data (they struggle even to imagine things in the left visual field).
这些都不是显而易见的结果。想象可能由一个独立于识别的系统执行。但在大脑皮层中，情况并非如此——它们在完全相同的区域执行。这正是我们对生成模型的期望：感知和想象不是独立的系统，而是同一枚硬币的两面。
None of this is an obvious result. Imagination could have been performed by a system separate from recognition. But in the neocortex, this is not the case—they are performed in the exact same area. This is exactly what we would expect from a generative model: perception and imagination are not separate systems but two sides of the same coin.
预测一切
Predicting Everything
思考大脑皮层中生成模型的一种方法是，它会模拟你的环境，这样它就可以在事情发生之前预测它们。大脑皮层会不断将实际感官数据与其模拟预测的数据进行比较。这就是你如何立即识别周围环境中发生的任何令人惊讶的事情。
One way to think about the generative model in the neocortex is that it renders a simulation of your environment so that it can predict things before they happen. The neocortex is continuously comparing the actual sensory data with the data predicted by its simulation. This is how you can immediately identify anything surprising that occurs in your surroundings.
当你走在街上时，你不会注意脚的感觉。但随着你的每一个动作，你的大脑皮层都会被动地预测它所期望的感觉结果。如果你把左脚放下来，感觉不到地面，你会立即看向看看你是否即将跌入坑洞。你的大脑皮层正在模拟你走路的情形，如果模拟结果与传感器数据一致，你就不会注意到，但如果它的预测是错误的，你就会注意到。
As you walk down the street, you are not paying attention to the feelings of your feet. But with every movement you make, your neocortex is passively predicting what sensory outcome it expects. If you placed your left foot down and didn’t feel the ground, you would immediately look to see if you were about to fall down a pothole. Your neocortex is running a simulation of you walking, and if the simulation is consistent with sensor data, you don’t notice it, but if its predictions are wrong, you do.
大脑从早期的两侧对称动物开始就一直在进行预测，但随着进化时间的推移，这些预测变得更加复杂。早期的两侧对称动物能够了解到一个神经元的激活往往先于另一个神经元的激活，因此可以利用第一个神经元来预测第二个神经元。这是最简单的预测形式。早期的脊椎动物可以利用世界上的模式来预测未来的奖励。这是一种更复杂的预测形式。早期的哺乳动物拥有大脑皮层，它们学会了预测不仅仅是反射的激活或未来的奖励；它们学会了预测一切。
Brains have been making predictions since early bilaterians, but over evolutionary time, these predictions became more sophisticated. Early bilaterians could learn that the activation of one neuron tended to precede the activation of another neuron and could thereby use the first neuron to predict the second. This was the simplest form of prediction. Early vertebrates could use patterns in the world to predict future rewards. This was a more sophisticated form of prediction. Early mammals, with the neocortex, learned to predict more than just the activation of reflexes or future rewards; they learned to predict everything.
大脑皮层似乎处于一种连续预测所有感官数据的状态。如果反射回路是反射预测机器，基底神经节中的批评者是奖励预测机器，那么大脑皮层就是世界预测机器——旨在重建动物周围的整个三维世界，以准确预测动物及其周围世界中的事物移动时接下来会发生什么。
The neocortex seems to be in a continuous state of predicting all its sensory data. If reflex circuits are reflex-prediction machines, and the critic in the basal ganglia is a reward-prediction machine, then the neocortex is a world-prediction machine—designed to reconstruct the entire three-dimensional world around an animal to predict exactly what will happen next as animals and things in their surrounding world move.
大脑皮层微电路以某种方式实现了这样一个通用系统，它可以模拟多种类型的输入。给它视觉输入，它将学会模拟世界的视觉方面；给它听觉输入，它将学会模拟世界的听觉方面。这就是为什么大脑皮层在任何地方看起来都一样。大脑皮层的不同子区域根据它们接收到的输入模拟外部世界的不同方面。把所有这些大脑皮层柱放在一起，它们就组成了一场模拟交响曲，呈现出一个丰富的三维世界，充满了可以看到、触摸到和听到的物体。
Somehow the neocortical microcircuit implements such a general system that it can render a simulation of many types of input. Give it visual input and it will learn to render a simulation of the visual aspects of the world; give it auditory input and it will learn to render a simulation of auditory aspects of the world. This is why the neocortex looks the same everywhere. Different subregions of neocortex simulate different aspects of the external world based on the input they receive. Put all these neocortical columns together, and they make a symphony of simulations that render a rich three-dimensional world filled with objects that can be seen, touched, and heard.
大脑皮层如何做到这一点仍是一个谜。至少有一种可能性是，它天生就具有做出一系列聪明假设的能力。现代人工智能模型通常被视为狭窄的——也就是说，它们只能在经过专门训练的狭窄情况下工作。而人类大脑则被认为是通用的——它能够在广泛的情况下工作。研究议程因此，我们一直在努力让人工智能更加通用。然而，我们可能把事情搞反了。新皮层之所以如此优秀，原因之一可能是，在某些方面，它远不如我们目前的人工神经网络那么通用。新皮层可能对世界做出明确的狭隘假设，而正是这些假设让它变得如此通用。
How the neocortex does this is still a mystery. At least one possibility is that it is prewired to make a set of clever assumptions. Modern AI models are often viewed as narrow—that is, they’re able to work in a narrow set of situations they are specifically trained for. The human brain is considered general—it is able to work in a broad set of situations. The research agenda has therefore been to try and make AI more general. However, we might have it backward. One of the reasons why the neocortex is so good at what it does may be that, in some ways, it is far less general than our current artificial neural networks. The neocortex may make explicit narrow assumptions about the world, and it may be exactly these assumptions that enable it to be so general.
预测的演变
The Evolution of Prediction
早期两侧对称动物的预测 PREDICTION IN EARLY BILATERIANS |
早期脊椎动物的预测 PREDICTION IN EARLY VERTEBRATES |
早期哺乳动物的预测 PREDICTION IN EARLY MAMMALS |
预测反射激活 Predict reflex activation |
预测未来奖励 Predict future rewards |
预测所有感官数据 Predict all sensory data |
反射电路 Reflex circuits |
皮质和基底神经节 Cortex and basal ganglia |
新皮质 Neocortex |
例如，大脑皮层可能预先假设传入的传感器数据（无论是视觉、听觉还是体感）代表独立于我们自身并可以自行移动的三维物体。因此，它不必了解空间、时间以及自我与他人之间的差异。相反，它试图通过假设它必须来自三维世界来解释它接收到的所有传入感官信息随着时间的推移而展开。
For example, the neocortex may be prewired to assume that incoming sensor data, whether visual, auditory, or somatosensory, represent three-dimensional objects that exist separately from ourselves and can move on their own. Therefore, it does not have to learn about space, time, and the difference between the self and others. Instead, it tries to explain all incoming sensory information it receives by assuming it must have been derived from a 3D world that unfolds over time.
这让我们对亥姆霍兹所说的推理有了些许的理解——大脑皮层中的生成模型试图推断其感官输入的原因。原因只是大脑皮层认为与所给感官输入最匹配的内部模拟三维世界。这也是为什么生成模型会试图解释其输入的原因——你的大脑皮层试图呈现一种可以产生你所看到的图像的世界状态（例如，如果那里有一只青蛙，它会“解释”为什么那些阴影看起来是那样的）。
This provides some intution about what Helmholtz meant by inference—the generative model in the neocortex tries to infer the causes of its sensory input. Causes are just the inner simulated 3D world that the neocortex believes best matches the sensory input it is being given. This is also why generative models are said to try to explain their input—your neocortex attempts to render a state of the world that could produce the picture that you are seeing (e.g., if a frog was there, it would “explain” why those shadows look the way they do).
但为什么要这样做呢？渲染外部世界的内部模拟有什么意义呢？新皮层为这些古代哺乳动物提供了什么价值？
But why do this? What is the point of rendering an inner simulation of the external world? What value did the neocortex offer these ancient mammals?
关于现代人工智能系统缺少什么以及如何让人工智能系统表现出人类水平的智能，目前存在许多争论。有些人认为，关键的缺失部分是语言和逻辑。但其他人，比如 Meta 的 AI 负责人 Yann LeCun，则认为它们是其他东西，更原始的东西，进化得更早的东西。用 LeCun 的话来说：
There are many ongoing debates about what is missing in modern AI systems and what it will take to get AI systems to exhibit human-level intelligence. Some believe the key missing pieces are language and logic. But others, like Yann LeCun, head of AI at Meta, believe they are something else, something more primitive, something that evolved much earlier. In LeCun’s words:
我们人类过于重视语言和符号作为智能的基础。灵长类动物、狗、猫、乌鸦、鹦鹉、章鱼和许多其他动物都没有类似人类的语言，但却表现出超越我们最好的人工智能系统的智能行为。它们所拥有的是学习强大的“世界模型”的能力，这种能力使它们能够预测其行为的后果，并搜索和计划行动以实现目标。学习这种世界模型的能力正是我们所缺少的来自当今的人工智能系统。
We humans give way too much importance to language and symbols as the substrate of intelligence. Primates, dogs, cats, crows, parrots, octopi, and many other animals don’t have human-like languages, yet exhibit intelligent behavior beyond that of our best AI systems. What they do have is an ability to learn powerful “world models” that allow them to predict the consequences of their actions and to search for and plan actions to achieve a goal. The ability to learn such world models is what’s missing from AI systems today.
哺乳动物大脑皮层（或许鸟类甚至章鱼也有类似的结构）呈现的模拟正是这个缺失的“世界模型”。大脑皮层如此强大的原因不仅在于它可以将其内部模拟与感官证据（亥姆霍兹的推理感知）相匹配，更重要的是，它的模拟可以独立探索。如果你有一个足够丰富的外部世界内部模型，你就可以在脑海中探索那个世界，预测你从未采取过的行为的后果。是的，你的大脑皮层使你能够睁开眼睛，认出你面前的椅子，但它也使你闭上眼睛，仍然能在脑海中看到那把椅子。你可以在脑海中旋转和修改椅子，改变它的颜色，改变它的材质。当你的大脑皮层中的模拟与你周围的真实外部世界脱钩时——当它想象出不存在的东西时——它的力量就变得最为明显。
The simulation rendered in the neocortices of mammals (and perhaps in similar structures of birds or even octopuses) is exactly this missing “world model.” The reason the neocortex is so powerful is not only that it can match its inner simulation to sensory evidence (Helmholtz’s perception by inference) but, more important, that its simulation can be independently explored. If you have a rich enough inner model of the external world, you can explore that world in your mind and predict the consequences of actions you have never taken. Yes, your neocortex enables you to open your eyes and recognize the chair in front of you, but it also enables you to close your eyes and still see that chair in your mind’s eye. You can rotate and modify the chair in your mind, change its colors, change its materials. It is when the simulation in your neocortex becomes decoupled from the real external world around you—when it imagines things that are not there—that its power becomes most evident.
这是大脑皮层赋予早期哺乳动物的礼物。想象力——能够呈现未来的可能性并重温过去事件的能力——是人类智力进化的第三次突破。从中出现了许多熟悉的智力特征，其中一些我们已经在人工智能系统中重新创造和超越，另一些则仍然超出我们的理解范围。但它们都是在第一批哺乳动物的微小大脑中进化而来的。
This was the gift the neocortex gave to early mammals. It was imagination—the ability to render future possibilities and relive past events—that was the third breakthrough in the evolution of human intelligence. From it emerged many familiar features of intelligence, some of which we have re-created and surpassed in AI systems, others of which are still beyond our grasp. But all of them evolved in the minuscule brains of the first mammals.
在接下来的章节中，我们将了解大脑皮层如何使早期哺乳动物能够执行计划、情景记忆和因果推理等技能。我们将了解如何利用这些技巧来实现精细运动技能。我们将了解大脑皮层如何实现注意力、工作记忆和自我控制。我们将看到，正是在早期哺乳动物的大脑皮层中，我们将发现许多类似人类的智能的秘密，这些秘密甚至在我们最聪明的人工智能系统中也找不到。
In the coming chapters, we will learn how the neocortex enabled early mammals to perform feats like planning, episodic memory, and causal reasoning. We will learn how these tricks were repurposed to enable fine motor skills. We will learn about how the neocortex implements attention, working memory, and self-control. We will see that it is in the neocortex of early mammals where we will find many of the secrets to human-like intelligence, those that are missing from even our smartest AI systems.
12
12
幻想秀中的老鼠
Mice in the Imaginarium
大脑皮层的出现是人类智力进化史上的一个分水岭。大脑皮层的原始功能肯定不如其现代应用那么广泛——它不是用来思考存在的本质、规划职业或写诗。相反，第一个大脑皮层赋予了早期哺乳动物更基础的东西：想象世界本来面目的能力。
THE EMERGENCE OF the neocortex was a watershed moment in the evolutionary history of human intelligence. The original function of the neocortex was surely not as broad as its modern applications—it wasn’t for pondering the nature of existence, planning careers, or writing poetry. Instead, the first neocortex gifted early mammals something more foundational: the ability to imagine the world as it is not.
大多数关于新皮质的研究都集中在其令人印象深刻的物体识别能力上：看到一张脸部照片，就能在多个尺度、平移和旋转下轻松识别它。在早期生成模型的背景下，生成模式——模拟过程——通常被视为实现识别优势的手段。换句话说，识别才是有用的；想象力是副产品。但在新皮质出现之前的皮质也能很好地识别物体；即使是鱼也能在旋转、缩放或扰动。
Most research on the neocortex has focused on its impressive ability to recognize objects: to see a single picture of a face and easily identify it at many scales, translations, and rotations. In the context of early generative models, the generative mode—the process of simulation—is often viewed as the means to achieving the benefit of recognition. In other words, recognition is what is useful; imagination is a byproduct. But the cortex that came before the neocortex could also recognize objects quite well; even fish can recognize objects when rotated, rescaled, or perturbed.
新皮层的核心进化功能可能恰恰相反——识别可能是解锁模拟适应优势的手段。这表明新皮层的原始进化功能不是识别世界（这是较老的脊椎动物皮层已经具备的能力），而是想象和模拟世界，这是较老的皮层所缺乏的能力。
The core evolutionary function of the neocortex might have been the opposite—recognition might have been the means that unlocked the adaptive benefit of simulating. This would suggest that the original evolutionary function of the neocortex was not recognizing the world—an ability the older vertebrate cortex already had—but instead imagining and simulating the world, an ability the older cortex was lacking.
新皮层模拟为早期哺乳动物提供了三种新能力，这三种能力对于在一亿五千万年的尖牙恐龙掠食攻击中生存下来至关重要。
There were three new abilities that neocortical simulating provided early mammals, all three of which were essential for surviving the one-hundred-and-fifty-million-year predatory onslaught of sharp-toothed dinosaurs.
新能力 #1：替代性尝试和错误
New Ability #1: Vicarious Trial and Error
20 世纪 30 年代，加州大学伯克利分校的心理学家爱德华·托尔曼 (Edward Tolman) 将老鼠放入迷宫中，观察它们的学习方式。当时，这属于正常的心理学工作——这是桑代克之后的一代。桑代克效果定律的研究范式（动物会重复产生愉快结果的行为）当时盛行一时。
In the 1930s, the psychologist Edward Tolman, working at UC Berkeley, was putting rats in mazes to see how they learned. Normal psychology-type work at the time—this was the generation that followed Thorndike. The research paradigm of Thorndike’s law of effect, in which animals repeated behavior that had pleasant consequences, was in full force.
托尔曼注意到了一件奇怪的事情。当老鼠到达迷宫中的岔路时，如果选择不明显——不清楚它们应该往左走还是往右走——老鼠会停下来，来回看几秒钟在选择方向之前，先思考一下。这在桑迪克的标准观点中是说不通的，他认为所有的学习都是通过反复试验而发生的——为什么暂停和来回摇头的行为会得到强化呢？
Tolman noticed something odd. When rats reached forks in his mazes where the choice was not obvious—where it wasn’t clear whether they should go left or right—rats would pause and look back and forth for a few seconds before choosing a direction. This made no sense in the standard Thorndikian view that all learning occurred through trial and error—why would the behavior of pausing and toggling one’s head back and forth have been reinforced?
托尔曼推测：老鼠在做出每个选择之前，会“反复试验”。托尔曼称之为“间接试错法”。
Tolman made a speculation: The rat was “playing out” each option before taking it. Tolman called this “vicarious trial and error.”
只有在难以做出决定时，老鼠才会表现出这种摇头行为。让决定变得困难的一种方法是让成本接近收益。假设你把一只老鼠放在一条隧道里，它会经过不同的门，每扇门都通向食物。假设当老鼠经过这些门时，会发出一种特定的声音，表示如果老鼠选择穿过门，它将需要等待多长时间才能得到食物。一种声音表示老鼠只需等待几秒钟；另一种声音表示老鼠需要等待半分钟。一旦老鼠了解了这一切是如何运作的，它们就不会在短暂延迟的门口摇头（它们会立即跑进去拿食物，大概在想，这显然值得等待）或在长时间延迟的门口摇头（它们会立即走过门：这不值得等待，我要去看看下一扇门）。但它们会在中等延迟的门口摇头（这值得等待吗，还是我应该等待？尝试下一个选项？）。
Rats expressed this head toggling behavior only when decisions were hard. One way to make decisions hard is for the costs to be close to the benefits. Suppose you put a rat in a tunnel where it passes various doors, each of which leads to food. And suppose that when rats pass these doors, a specific sound is made that signals how long a rat will have to wait for food if it chooses to walk through the door. One sound signals that rats will have to wait only a few seconds; another sound signals that the rat will have to wait up to half a minute. Once rats learn how all this works, they do not perform their head toggling at the doors with a short delay (they just immediately run in to get the food, presumably thinking, Obviously this is worth the wait) or at doors with a long delay (they just immediately walk past the door: No way this is worth the wait, I’ll just check out the next door). But they do perform the head toggling behavior at the doors with a medium delay (Is this worth the wait or should I try the next option?).
让决策变得困难的另一种方法是改变规则。如果食物突然不再出现在迷宫中老鼠预期的位置，那么下次将老鼠放入迷宫时，它的头部切换行为就会大大增加，似乎考虑替代路径。类似地，假设一只老鼠进入一个有两种食物的迷宫，假设最近这只老鼠吃了很多这类食物（因此不再想吃这种食物），然后这种头部切换行为就会出现（我想左转去拿 X 而不是右转得到 Y？）。
Another way to make decisions difficult is to change the rules. If suddenly food is no longer where a rat expects it in the maze, the next time the rat is placed in the maze it will substantially increase its head toggling behavior, seeming to consider alternative paths. Similarly, suppose a rat enters a maze with two types of food, and suppose recently the rat has eaten a lot of one of those types of food (and hence no longer wants that food), then this head toggling behavior emerges (Do I want to turn left to get X instead of turning right and getting Y?).
当然，你观察到老鼠停顿并来回转动头部这一事实并不能证明它实际上在想象着走不同的路。由于缺乏证据，在托尔曼的观察之后的几十年里，间接试错法的想法逐渐失宠。直到最近，在 21 世纪，技术才发展到可以实时记录老鼠大脑中神经元群在环境中导航时的程度。这是神经科学家第一次能够真正观察老鼠停顿并来回摆动头部时大脑中发生的事情。
Of course, the fact that you can observe a rat pause and turn its head back and forth doesn’t prove that it is, in fact, imagining going down different paths. And due to this lack of evidence, the idea of vicarious trial and error faded out of favor in the decades that followed Tolman’s observation. It was only recently, in the 2000s, that technology reached the point where ensembles of neurons in the brain of a rat could be recorded in real time as rats navigated their environment. This was the first time that neuroscientists could literally watch what was happening in the brains of rats when they paused and toggled their heads back and forth.
明尼苏达大学的神经科学家戴维·雷迪什和他的学生亚当·约翰逊首次探究了老鼠在做出这些选择时大脑中发生了什么。当时，众所周知，当老鼠在迷宫中穿行时，其海马体中的特定位置细胞会被激活。这类似于鱼的空间地图——特定的海马神经元编码特定位置。在鱼中，只有当鱼实际出现在编码位置时，这些神经元才会活跃——但当雷迪什和约翰逊在老鼠身上记录这些神经元时，他们发现了不同的东西：当老鼠在决策点停下来并来回转动头部时，它的海马体停止编码老鼠的实际位置，而是来回快速播放位置代码序列，这些序列构成了从选择点开始的两条可能的未来路径。雷迪什可以真正看到老鼠想象未来的路径。
It was David Redish and his student Adam Johnson, neuroscientists at the University of Minnesota, who first probed what was happening in the brain of rats during these choices. At the time, it was well known that when a rat navigates a maze, specific place cells in its hippocampus get activated. This was similar to the spatial map of a fish—specific hippocampal neurons encode specific locations. In a fish, these neurons become active only when the fish is physically present at the encoded location—but when Redish and Johnson recorded these neurons in rats, they found something different: when the rat stopped at the decision point and turned its head back and forth, its hippocampus ceased to encode the actual location of the rat and instead went back and forth rapidly playing out the sequence of place codes that made up both possible future paths from the choice point. Redish could literally see the rat imagining future paths.
这项技术的突破性无论怎样强调都不为过——神经科学家直接观察老鼠的大脑，直接观察老鼠思考各种未来。托尔曼是对的：他观察到的头部转动行为确实是老鼠在计划未来的行动。
How groundbreaking this was cannot be overstated—neuroscientists were peering directly into the brain of a rat, and directly observing the rat considering alternative futures. Tolman was right: the head toggling behavior he observed was indeed rats planning their future actions.
相比之下，第一批脊椎动物并没有提前计划自己的行动。我们可以通过研究它们的冷血后代——现代鱼类和爬行动物——来发现这一点，它们没有表现出通过间接试错进行学习的证据。
In contrast, the first vertebrates did not plan their actions ahead of time. We can see this by examining their cold-blooded descendants—modern fish and reptiles—who show no evidence of learning by vicarious trial and error.
考虑绕行任务。拿一条鱼，把它放在一个中间有透明屏障的鱼缸里。在鱼缸的一角打一个小洞。把屏障打开，这样鱼就可以从一边游到另一边。让鱼探索鱼缸，找到洞，花些时间来回游动。几天后，做一些新的事情：把鱼放在鱼缸的一侧，把食物放在透明屏障的另一侧。会发生什么？
Consider the detour task. Take a fish and put it in a tank with a transparent barrier in the middle of the tank. Put a small hole in one corner of the barrier so that the fish can get from one side to the other. Let the fish explore the tank, find the hole, and spend some time swimming back and forth. After several days, do something new: Put the fish on one side of the tank and put a treat on the opposite side of the transparent barrier. What happens?
如果你想要食物，那么最聪明的做法是立即从食物处游到屏障的角落，穿过洞口，然后转回来在另一边吃食物。但这不是鱼会做的事情。鱼会直接撞到透明屏障上去吃食物。在撞到墙上足够多次后，它放弃了，继续在环境中游荡。最终，在游荡过程中，鱼碰巧再次穿过洞口，但即使在这里，它也没有意识到现在可以吃到食物了；鱼没有转向食物。相反，鱼只是继续穿过鱼缸的另一边。只有当它碰巧转身再次看到食物时，它才会兴奋地冲向食物。事实上，对于一条有过多次在屏障两侧之间穿行经验的鱼来说，找到食物所需的时间与一条从未经历过到达另一侧的鱼一样多。没有任何障碍。
The smart thing to do, if you wanted food, would be to immediately swim away from the treat to the corner of the barrier, go through the opening, and then turn back to get the food on the other side. But this is not what the fish does. The fish runs directly into the transparent barrier trying to get the food. After bopping against the wall enough times, it gives up and continues wandering its environment. Eventually, while wandering the environment, the fish happens to pass through the hole again, but even here it shows no understanding that it can now access the food; the fish does not turn toward the food. Instead, the fish just continues onward through the other side of the tank. Only when it happens to turn back and see the food again does it excitedly dart toward it. Indeed, it takes the same amount of time to find the food for a fish who has had many experiences navigating between each side of the barrier, as it does for a fish that has never experienced getting to the other side of the barrier at all.
这是为什么呢？虽然鱼以前也曾游过洞到达鱼缸的另一边，但它从未知道穿过洞的路径会产生多巴胺。反复试验的学习从未训练过鱼的基底神经节，当它看到透明屏障对面的食物时，它应该游过洞去获取食物。
Why is this? While the fish had swum through the hole to get to the other side of the tank before, it had never learned that the path through the hole provided dopamine. Trial-and-error learning had never trained the fish’s basal ganglia that when it saw food across the transparent barrier, it should take the action of swimming through the hole to get the food.
这是边做边学的一个关键问题：虽然鱼已经学会了穿过洞的路径，但它以前从未走过那条路去觅食。所以当它看到食物时，它所能做的就是直接向食物发出“接近”信号。然而，老鼠要精明得多。在这种绕行任务中，它们的表现远远胜过鱼。老鼠和鱼最初都会跑到透明屏障前试图获取食物。但老鼠更善于弄清楚如何绕过障碍物。一只探索过地图的老鼠——知道如何到达透明障碍物的另一边（即使这样做从未得到奖励）——会比一只从未绕过障碍物的老鼠更快地到达另一边。这揭示了替代性尝试的好处之一，错误：一旦老鼠对其周围环境有了世界模型，它们就可以快速地在脑海中探索它，直到找到绕过障碍物来得到它们想要的东西的方法。
This is a critical problem of learning only by doing: Although the fish had learned the path through the hole, it had never taken that path to get food before. So when it saw food, all it could do was generate an “approach” signal directly toward the food. Rats, however, are much savvier. In such detour tasks, they dramatically outperform fish. Both rats and fish will initially run up to the transparent barrier to try and get food. But rats are much better at figuring out how to navigate around a barrier. And a rat who has explored a map well—who knows how to get to the other side of a transparent barrier (even if doing so was never rewarded)—will get to the other side much faster than will a rat who has never navigated around the barrier before. This reveals one of the benefits of vicarious trial and error: once a rat has a world model of their environment, they can rapidly mentally explore it until they find a way to get around obstacles to get what they want.
旧的边做边学策略的另一个问题是，有时过去的奖励不能预测当前的奖励，因为动物的内部状态已经发生了变化。例如，把一只老鼠放在一个迷宫里，迷宫的一边提供过咸的食物，另一边提供正常的食物。让老鼠正常地穿过这个迷宫，并尝试过咸的食物（它会讨厌然后避开）和正常的食物（老鼠会喜欢）。现在假设你把老鼠放回那个情境中，但有一个变化：你让它严重缺盐。老鼠会怎么做？
Another problem with the older strategy of learning by doing is that sometimes past rewards are not predictive of current rewards because an animal’s internal state has changed. For example, put a rat in a maze where one side provides overly salty food and the other side provides normal food. Let the rat navigate this maze normally and try the overly salty food (which it will hate and then avoid) and the normal food (which the rat will enjoy). Now suppose you put the rat back in that situation but with a twist: You make it severely salt-deprived. What does the rat do?
它立即 跑向盐。这很了不起，因为老鼠现在跑向迷宫中之前受到负面强化的部分。这之所以可能，只是因为老鼠“模拟”了每条路径，并间接意识到过咸的食物现在会带来巨大的回报。换句话说，在老鼠采取行动之前，通往盐的路径就得到了间接强化。我并不知道有任何研究表明鱼或爬行动物可以完成这样的任务。
It immediately runs toward the salt. This is remarkable because the rat is now running toward a part of the maze that was previously negatively reinforced. This is possible only because the rat “simulated” each path and realized—vicariously—that the overly salty food would now be hugely rewarding. In other words, the path toward salt was vicariously reinforced before the rat even acted. I am unaware of any studies showing that fish or reptiles can perform such a task.
新能力 #2：反事实学习
New Ability #2: Counterfactual Learning
人类会花很多时间沉浸在悔恨中。在普通人的谈话中，你可能会听到这样的问题：“当拉米雷斯提出放弃我们的生活，搬到智利去他的农场工作时，如果我答应了，生活会怎样？”“如果我追随自己的梦想，追求棒球而不是这份办公室工作，生活会怎样？”“我今天为什么在工作中说了那句蠢话？如果我说了更聪明的话，会发生什么？”
Humans spend a painful amount of time wallowing in regret. Questions you might hear from an average human conversation: “What would life have been like if I had said yes when Ramirez offered to throw our lives away and move to Chile to work on his farm?” “What if I had followed my dream and pursued baseball instead of this desk job?” “Why did I say that stupid thing at work today? What would have happened had I said something cleverer?”
佛教徒和心理学家都意识到，反复思考本可以发生的事情是人类痛苦的根源。我们无法改变过去，那么为什么要用它来折磨自己呢？这种现象的进化根源可以追溯到早期哺乳动物。在古代世界以及后来的许多世界里，这种反复思考是有用的，因为同样的情况经常会重现，人们可以做出更好的选择。
Buddhists and psychologists alike realize that ruminating about what could have been is a source of great misery for humanity. We cannot change the past, so why torture ourselves with it? The evolutionary roots of this go back to early mammals. In the ancient world, and in much of the world that followed, such ruminating was useful because often the same situation would recur and a better choice could be made.
我们在早期脊椎动物中看到的强化学习类型缺陷：它只能强化实际采取的具体行动。这种策略的问题在于，实际采取的路径只是所有可能采取的路径的一小部分。动物第一次尝试选择最佳路径的可能性有多大？
The type of reinforcement learning we saw in early vertebrates has a flaw: It can only reinforce the specific action actually taken. The problem with this strategy is that the paths that were actually taken are a small subset of all the possible paths that could have been taken. What are the chances that an animal’s first attempt picked the best path?
如果一条鱼游进鱼群捕食一些无脊椎动物并带走一只，而附近的另一条鱼选择了不同的路径并带走四只，那么第一条鱼不会从错误中吸取教训；它只会用平庸的奖励来强化它所走的路径。鱼所缺乏的是从反事实中学习的能力。反事实是如果你过去做出了不同的选择，世界现在会是什么样子。
If a fish swims into a shoal to hunt some invertebrates and comes away with one, and a nearby fellow fish took a different path and came away with four, the first fish won’t learn from that mistake; it will merely reinforce the path taken with the mediocre reward. What fish are missing is the ability to learn from counterfactuals. A counterfactual is what the world would be now if you had made a different choice in the past.
大卫·雷迪什发现老鼠可以想象不同的未来，于是想看看老鼠是否也能想象不同的过去选择。雷迪什和他的学生亚当·施泰纳把老鼠放在一个圆形迷宫里他们给老鼠设置了一条他们称之为“餐厅行”的走廊。老鼠逆时针跑来跑去，不断经过相同的四条走廊。每条走廊的尽头都有不同的“餐厅”，里面有不同口味的食物（巧克力、樱桃、香蕉）。当老鼠经过每条走廊时，会发出一个随机的音调，表示如果老鼠在这条走廊等待而不是继续前往下一条走廊，食物会在等待时间后发放。音调 A 表示如果老鼠在当前走廊等待，它将在 45 秒内得到食物；音调 B意味着它将在五秒钟内得到食物。如果老鼠选择不等待，那么一旦它到达下一个走廊，它就不能再回去了；除非它再次绕着圆圈走一圈，否则食物不会发放。这给老鼠带来了一系列不可逆转的选择。老鼠有一小时的时间继续尝试尽可能多地获取它们最喜欢的食物。
David Redish, having discovered that rats can imagine alternative futures, wanted to see if rats could also imagine alternative past choices. Redish and one of his students Adam Steiner put rats in a circular maze they called “restaurant row.” Rats ran around counterclockwise and kept passing the same four corridors. At the end of each corridor was a different “restaurant,” containing a different flavor of food (chocolate, cherry, banana). As the rat passed each corridor, a random tone was sounded that signaled the delay before the food would be released if the rat waited at this corridor instead of continuing on to the next. Tone A meant if the rat waited at the current corridor, it would get the food within forty-five seconds; tone B meant it would get the food in five seconds. If a rat chose not to wait, then once it got to the next corridor, it could no longer go back; the food would not be released unless it went all the way around the circle again. This presented rats with a continuous set of irreversible choices. Rats were given an hour to keep trying to get as much of their favorite foods as they could.
图 12.1：老鼠的 Redish 餐厅行后悔测试
Figure 12.1: Redish restaurant-row test of regret in rats
原始人物模型由 Max Bennett 绘制，由 David Redish 监督并授权
Original figure by Max Bennett, with oversight and permission from David Redish
考虑一下在特定走廊上老鼠面临的选择：我是在这里等待普通的香蕉（音调刚刚发出信号，五秒钟后就会发出），还是跑到下一扇门（里面有我最喜欢的食物樱桃），赌一赌它也会很快被释放？当老鼠选择放弃快速获得香蕉的机会而去尝试樱桃门，而下一个音调又发出信号，表示要等待四十五秒时，老鼠表现出后悔选择的所有迹象。它们停下来，回头看向它们已经经过的走廊，再也无法回去了。大脑皮层味觉区域的神经元重新激活了香蕉的表征，表明老鼠实际上在想象一个它们做出了不同选择并吃到香蕉的世界。
Consider the choice presented to the rats at a given corridor: Do I wait here for mediocre banana that the tone just signaled would be released in five seconds, or do I run to the next door, which contains my favorite food, cherry, and gamble that it will also be released quickly? When rats chose to forgo quick access to a banana treat to try the cherry door and the next tone signaled a long wait of forty-five seconds, rats showed all the signs of regretting their choice. They paused and looked back toward the corridor that they had passed and could no longer go back to. And the neurons in the taste area of the neocortex reactivated the representation of banana, showing that rats were literally imagining a world where they had made a different choice and got to eat the banana.
那些返回并重新激活放弃选择的表征的老鼠最终也改变了它们未来的选择。它们下一次等待的时间更长，并且更匆忙地吃其他食物，试图回到迷宫周围的樱桃处再次尝试。
The rats that turned back and reactivated the representation of the forgone choice also ended up changing their future choices. They waited longer the next time and ate other food more hastily to try and get back around the maze to the cherry to try again.
灵长类动物也会推理反事实。考虑一个实验，猴子被教导玩石头、剪刀、布。每当猴子输了，它们的下一步总是偏向于在前一场比赛中获胜的举动。如果一只猴子输了，因为它选择了布而它的对手选择了剪刀，那么下一次它最有可能选择石头（这会赢得剪刀）。这无法用早期脊椎动物的时间差异学习来解释。根据这里提出的进化框架，如果一条鱼可以玩石头、剪刀、布，它就不会表现出这种效果。如果一条鱼在用布对抗剪刀后输了，它下次出布的可能性就会降低（该行为受到惩罚，因为它会导致输掉比赛），但鱼下次出石头或剪刀的可能性是相等的（这两种行为都不会受到惩罚或强化）。相比之下，因为猴子可以通过出布来重演输掉比赛后可能发生的事情，它会意识到，如果它出石头，它就会赢了。因此，猴子们根据反事实学习改变了自己的行为。
Primates also reason about counterfactuals. Consider an experiment where monkeys were taught to play rock, paper, scissors. Whenever monkeys lost, their next move was always biased toward the move that would have won in the prior game. If a monkey lost because it chose paper while its opponent chose scissors, next time it is most likely to pick rock (which would have won against scissors). This cannot be explained with the temporal difference learning of early vertebrates. According to the evolutionary framework presented here, if a fish could play rock, paper, scissors, it would not show this effect. If a fish lost after playing paper against scissors, it would become less likely to play paper next time (that action was punished because it led to losing), but the fish would be equally likely to play rock or scissors next time (neither action was punished or reinforced). In contrast, because a monkey can replay what could have been after losing by playing paper, it will realize that if it had played rock it would have won. Thus, monkeys are changing their actions based on counterfactual learning.
因果关系的认知可能与反事实学习的概念密切相关。当我们说“X 导致 Y”时，我们的意思是，在 X 没有发生的反事实情况下，Y也没有发生。这就是我们区分相关性和因果关系的方法。如果你看到闪电击中一片干燥的森林，然后立刻起火，你会说是闪电引起了火灾，而不是火灾引起了闪电。你之所以这样说，是因为当你想象反事实的情况，即闪电没有击中时，火灾就不会发生。没有反事实，就无法区分因果关系和相关性。你永远不知道什么导致了什么；你只能知道“X 总是在 Y 之前发生”或“每当 X 发生时，Y 就会发生”或“没有 X 发生，Y 就不会发生”，等等。
The perception of causation may be intricately tied to the notion of counterfactual learning. What we mean when we say “X caused Y” is that in the counterfactual case where X did not occur, then Y did not occur either. This is how we tease out the difference between correlation and causation. If you saw lightning strike a dry forest, and a fire immediately started, you would say that it was the lightning that caused the fire, not the fire that caused the lightning. You say this because when you imagine the counterfactual case in which lightning did not strike, the fire did not appear. Without counterfactuals, there is no way to distinguish between causation and correlation. You can never know what caused what; you can know only that “X always happens before Y” or “Whenever X happens, Y happens” or “Y has never happened without X happening,” and so on.
反事实学习代表了祖先大脑在解决信用分配问题方面取得的重大进步。提醒一下，信用分配问题是这样的：当发生一些重要事件时，您希望能够提前预测，您如何选择哪些先前的行为或事件可以预测该事件？简而言之：发生了许多事情（鸟鸣、风吹、树叶摇晃和闪电），然后发生了火灾——您认为什么可以很好地预测火灾？在早期的两侧对称动物中，阻挡、潜在抑制和遮蔽等简单技巧驱动着做出简单预测和联想的逻辑。在早期脊椎动物中，时间差异学习的进化使基底神经节能够使用未来预测奖励的变化来分配信用；当批评者认为情况变得更好或更糟时，线索或行动就会得到赞扬。但在早期哺乳动物中，由于它们能够模拟替代的过去，因此也可以用反事实来分配信用。通过询问“如果我不采取这一举动，我会输掉比赛吗？”，哺乳动物可以判断这一举动是否真正值得赢得比赛。
Counterfactual learning represented a major advancement in how ancestral brains solved the credit assignment problem. As a reminder, the credit assignment problem is this: When some important event occurs that you want to be able to predict ahead of time, how do you choose what previous actions or events to give credit for having been predictive of the event? In simple terms: A bunch of things happen (a bird chirps, a gust of wind blows, a leaf moves, and lightning strikes), and then a fire appears—what do you give credit for being a good predictor of fire? In early bilaterians, simple tricks like blocking, latent inhibition, and overshadowing drove the logic by which simple predictions and associations were made. In early vertebrates, the evolution of temporal difference learning enabled the basal ganglia to assign credit using changes in future predicted reward; when the critic thinks the situation just got better or worse is when cues or actions are given credit. But in early mammals, with their ability to simulate alternative pasts, credit could also be assigned with the counterfactual. By asking “Would I have lost the game had I not made this move?,” mammals can determine whether a move truly deserves credit for winning the game.
因果关系本身可能更多地存在于心理学而非物理学中。没有实验可以明确证明因果关系的存在；它是完全无法衡量的。我们进行的受控实验可能表明因果关系，但它们总是无法证明，因为你实际上永远无法进行完全控制的实验。因果关系即使是真实存在的，在经验上也总是遥不可及的。事实上，现代量子力学领域的实验表明，因果关系可能根本不存在，至少不是在任何地方都存在。物理定律可能包含现实特征如何从一个时间步骤发展到下一个时间步骤的规则，而事物之间根本没有任何真正的因果关系。最终，不管因果关系是否真实，我们对因果关系的直觉感知的演变不是源于它的真实性，而是源于它的有用性。因果关系是由我们的大脑构建的，使我们能够从过去的其他选择中间接学习。
Causation itself may live more in psychology than in physics. There is no experiment that can definitively prove the presence of causality; it is entirely immeasurable. Controlled experiments we run may suggest causation, but they always fall short of proof because you can, in fact, never run a perfectly controlled experiment. Causation, even if real, is always empirically out of reach. In fact, modern experiments in the field of quantum mechanics suggest that causation may not even exist, at least not everywhere. The laws of physics may contain rules of how features of reality progress from one time step to the next without any real causal relationships between things at all. Ultimately, whether causality is real or not, the evolution of our intuitive perception of causality does not derive from its reality but from its usefulness. Causation is constructed by our brains to enable us to learn vicariously from alternative past choices.
信用分配的演变
The Evolution of Credit Assignment
早期两侧对称动物的功劳分配 CREDIT ASSIGNMENT IN EARLY BILATERIANS |
早期脊椎动物的学分分配 CREDIT ASSIGNMENT IN EARLY VERTEBRATES |
早期哺乳动物的功劳分配 CREDIT ASSIGNMENT IN EARLY MAMMALS |
根据阻断、潜在抑制和遮蔽的基本规则分配信用 Credit assigned based on basic rules of blocking, latent inhibition, and overshadowing |
根据评论家预测未来奖励的变化来分配信用 Credit assigned based on when the critic predicts changes in future rewards |
基于反事实分配信用——如果先前的事件或行动没有发生，是否会阻止后续事件的发生（即，真正导致该事件的原因是什么？） Credit assigned based on the counterfactual–which previous events or actions, if they had not occurred, would have prevented the subsequent event from occurring (i.e., what truly caused the event?) |
虽然这种反事实学习和因果推理确实存在于许多哺乳动物，甚至是老鼠身上，但没有令人信服的证据表明鱼类或爬行动物能够从反事实中学习或推理因果关系（不过，有证据表明鸟类的推理能力）。这表明，至少在我们的血统中，这种能力首先出现在我们的哺乳动物祖先中。
While such counterfactual learning and causal reasoning is indeed seen across many mammals, even rats, there is no convincing evidence that fish or reptiles are capable of learning from counterfactuals or reasoning about causation (however, there is evidence of this type of reasoning in birds). This suggests that, at least in our lineage, this ability first emerged in our mammalian ancestors.
新能力#3：情景记忆
New Ability #3: Episodic Memory
1953 年 9 月，一位名叫亨利·莫莱森的 27 岁男子接受了一项实验性手术，切除了他的整个海马体——这是他衰弱性癫痫发作的根源。从某种意义上说，这次手术是成功的：他的癫痫发作严重程度明显减轻，他保留了他的个性和才智得到了恢复。但他的医生很快发现，手术夺走了病人最宝贵的东西：醒来后，莫莱森完全无法产生新的记忆。他可以进行一两分钟的对话，但不久之后就会忘记刚刚发生的一切。即使四十年后，他可以用 1953 年之前的事实来填字游戏，但不能用 1953 年之后发生的事实来填字游戏。莫莱森被困在 1953 年。
In September 1953, a twenty-seven year old man named Henry Molaison underwent an experimental procedure that removed his entire hippocampus—the source of his debilitating seizures. The surgery was, in one sense, a success: severity of his seizures were markedly reduced and he retained his personality and intellect. But his doctors quickly discovered that the surgery had deprived their patient of something precious: upon waking up, Molaison was entirely unable to produce new memories. He could hold a conversation for a minute or two, but shortly thereafter would forget everything that just happened. Even forty years later, he could fill out crosswords with facts from before 1953, but not with facts that occurred after. Molaison was stuck in 1953.
我们回顾过去不只是为了考虑其他过去的选择；我们回顾过去也是为了记住以前的生活事件。人们可以很容易地回忆起五分钟前他们做了什么，他们在大学主修什么，或者在婚礼致辞时讲的那个有趣的笑话。这种记忆形式，即我们回忆生活中特定的过去情节，被称为“情景记忆”。这与程序记忆不同，在程序记忆中，我们记得如何做各种动作，比如说话、打字或投掷棒球。
We don’t review the past only for the purpose of considering alternative past choices; we also review the past to remember previous life events. People can easily recall what they did five minutes ago or what they majored in at college or that funny joke made during a wedding speech. This form of memory, in which we recall specific past episodes of our lives, is called “episodic memory.” This is distinct from, say, procedural memory, where we remember how to do various movements, such as speaking, typing, or throwing a baseball.
但奇怪的是，我们并不真正记得情景事件。情景记忆的过程是模拟对过去的近似再现。当你想象未来事件时，你是在模拟未来的现实；当你回忆过去事件时，你是在模拟过去的现实。两者都是模拟。
But here is the weird thing—we don’t truly remember episodic events. The process of episodic remembering is one of simulating an approximate re-creation of the past. When imagining future events, you are simulating a future reality; when remembering past events, you are simulating a past reality. Both are simulations.
我们基于两个证据知道这一点。首先，回忆过去的事件和想象未来的事件使用的神经回路相似，甚至相同。当你想象未来时，激活的网络与想象未来时激活的网络相同。你回忆的是过去。回忆具体的东西（面孔、房屋）会重新激活感觉新皮层中的相同神经元，就像你实际感知这些事物时一样（就像我们用想象去看事物。
We know this based on two pieces of evidence. First, remembering past events and imagining future events use similar if not the same neural circuitry. The same networks get activated when you’re imagining the future as when you’re remembering the past. Remembering specific things (faces, houses) reactivates the same neurons in the sensory neocortex as when you actually perceive those things (just as we saw with imagining things).
情景记忆只是一种模拟的第二个证据来自与此类记忆相关的现象。例如，事实证明，人们的情景记忆是在记忆过程中“填充”的（很像视觉形状的填充）。这就是为什么情景记忆感觉如此真实，但准确性却远不如我们想象的那样。目击证人的证词最能证明情景记忆的缺陷。看着一排潜在罪犯的人，通常 100% 确定哪个人犯了罪。然而，与我们对记忆准确性的认知相反，它事实证明，目击证人的证词非常糟糕：77% 被“清白计划”平反的被错误定罪的人最初被定罪的原因是错误的目击证词。在大脑皮层中，虚构的想象场景和真实的情景记忆之间的区别很模糊；研究表明，反复想象一个过去没有发生过的错误事件，会增强一个人的信心，事件确实发生了。
The second piece of evidence that episodic memory is just a simulation comes from phenomena associated with such memories. For example, it turns out people’s episodic memories are “filled in” during the remembering process (much like the filling in of visual shapes). This is why episodic memories feel so real but are much less accurate than we think. The clearest demonstrations of episodic memory’s flaws are in eyewitness testimonies. Someone looking at a lineup of potential criminals is often 100 percent sure which member of the lineup committed the crime. However, contrary to our perceptions of how accurate our memories are, it has been shown that eyewitness testimony is terrible: 77 percent of the wrongfully convicted individuals exonerated by the Innocence Project were originally convicted because of mistaken eyewitness testimony. The distinction between a made-up imagined scene and an actual episodic memory is thin in the neocortex; studies show that repeatedly imagining a past event that did not occur falsely increases a person’s confidence that the event did occur.
可以通过向动物询问关于最近事件的意外问题来测试其情景记忆的存在。例如，老鼠可能会知道，当面对某个迷宫时，如果且仅当它在前几分钟遇到食物，它才会沿着一条路径找到食物；否则，它应该沿着另一条路径寻找食物。这个问题是随机提出的（通过随机向老鼠展示这个迷宫），因此很难看出第一批脊椎动物的 TD 学习机制如何学习这种偶然性；迷宫与每个方向的配对相同。然而，当老鼠被放在这个迷宫中时，它们很容易学会回忆它们最近是否遇到过食物，然后选择正确的对应路径为了获得更多的食物。目前唯一被证实具有这种情景记忆的非哺乳动物是鸟类和头足类动物，这两类物种似乎独立进化出了自己的大脑用于渲染模拟的结构。
One can test the presence of episodic memory in animals by asking them unexpected questions about recent events. For example, a rat may learn that when presented with a certain maze, it will find food down one path if and only if it encountered food in the prior few minutes; otherwise, it should go down a different path to get food. The fact that this question is asked randomly (by randomly presenting the rat with this maze) makes it hard to see how the TD-learning mechanisms of the first vertebrates would learn this contingency; the maze is equally paired with each direction. And yet rats easily learn, when placed in this maze, to recall whether they recently encountered food and then choose the right corresponding path to get more food. The only nonmammals that have been demonstrated to have such episodic memory are birds and cephalopods, the two groups of species that seem to have independently evolved their own brain structures for rendering simulations.
莫莱森手术后，他成为历史上研究最多的神经科学患者：为什么海马体是创建新情景记忆所必需的，而不是检索旧记忆所必需的？这是进化将旧结构重新用于新用途的一个例子。在哺乳动物的大脑中，情景记忆来自较老的海马体和较新的大脑皮层之间的合作。海马体可以快速学习模式，但无法呈现对世界的模拟；大脑皮层可以模拟世界的细节，但不能快速学习新的模式。情景记忆必须快速存储，因此，专为快速识别地点模式而设计的海马体被重新用于帮助快速编码情景记忆。通过重新激活海马体中的相应模式，可以“检索”感觉大脑皮层的分布式神经激活（即模拟）。就像老鼠重新激活海马体中的位置细胞来模拟走不同的路线一样路径，老鼠可以重新激活海马体中的这些“记忆代码”，以重新呈现最近事件的模拟。
After Molaison’s surgery, he became the most studied neuroscience patient in history: why was the hippocampus required for creating new episodic memories, but not for retrieving old ones? This is an example of evolution repurposing old structures for new purposes. In mammal brains, episodic memory emerges from a partnership between the older hippocampus and the newer neocortex. The hippocampus can quickly learn patterns, but cannot render a simulation of the world; the neocortex can simulate detailed aspects of the world, but cannot learn new patterns quickly. Episodic memories must be stored rapidly, and thus the hippocampus, designed for the rapid pattern recognition of places, was repurposed to also aid in the rapid encoding of episodic memories. Distributed neural activations of sensory neocortex (i.e., simulations) can be “retrieved” by reactivating the corresponding pattern in the hippocampus. Just as rats reactivated place cells in the hippocampus to simulate going down different paths, rats can reactivate these “memory codes” in the hippocampus to rerender simulations of recent events.
这种动态为灾难性遗忘问题提供了一种新的解决方案，即神经网络在学习新模式时会忘记旧模式。通过检索和重放近期记忆以及旧记忆，海马体帮助大脑皮层整合新记忆而不会破坏旧记忆。在人工智能中，这个过程被称为“生成重放”或“经验重放”，并且已被证明是解决灾难性遗忘的有效方法。这就是为什么海马体对于创建新记忆是必需的，但对于检索旧记忆却不是；大脑皮层可以在足够数量的重放后自行检索记忆。
This dynamic provided a new solution to the catastrophic forgetting problem, whereby neural networks forget old patterns when they learn new ones. By retrieving and replaying recent memories alongside old memories, the hippocampus aids the neocortex in incorporating new memories without disrupting old ones. In AI, this process is called “generative replay” or “experience replay” and has been shown to be an effective solution to catastrophic forgetting. This is why the hippocampus is necessary for creating new memories, but not for retrieving old ones; the neocortex can retrieve memories on its own after a sufficient amount of replay.
所有这些对未来和过去的模拟在机器学习中都有更大的类似之处。我们在突破 #2中看到的强化学习类型——时间差分学习——是一种无模型强化学习。在这种强化学习中，人工智能系统通过直接关联刺激、动作和奖励来学习。这些系统被称为“无模型”，因为它们不需要模型在做出决定之前展示可能的未来行动。虽然这使得 TD 学习系统高效，但也降低了它们的灵活性。
All this simulating of futures and pasts has a larger analog in machine learning. The type of reinforcement learning—temporal difference learning—that we saw with breakthrough #2 is a form of model-free reinforcement learning. In this type of reinforcement learning, AI systems learn by making direct associations between stimuli, actions, and rewards. These systems are called “model-free” because they do not require a model to play out possible future actions before making a decision. While this makes TD learning systems efficient, it also makes them less flexible.
还有另一类强化学习，称为基于模型的 强化学习。这些系统必须学习一些更复杂的东西：他们的行为如何影响世界的模型。一旦建立了这样的模型，这些系统就会在做出选择之前执行一系列可能的操作。这些系统更加灵活，但在做决定时，它们承担着构建和探索内部世界模型的艰巨任务。
There is another category of reinforcement learning called model-based reinforcement learning. These systems must learn something more complicated: a model of how their actions affect the world. Once such a model is constructed, these systems then play out sequences of possible actions before making choices. These systems are more flexible but are burdened with the difficult task of building and exploring an inner world model when making decisions.
现代技术中采用的大多数强化学习模型都是无模型的。掌握各种 Atari 游戏的著名算法游戏和许多自动驾驶汽车算法是无模型的。这些系统不会停下来考虑自己的选择；它们会立即根据所获得的感官数据采取行动。
Most of the reinforcement learning models employed in modern technology are model-free. The famous algorithms that mastered various Atari games and many self-driving-car algorithms are model-free. These systems don’t pause and consider their choices; they immediately act in response to the sensory data they are given.
事实证明，基于模型的强化学习更难实现，原因有二。
Model-based reinforcement learning has proven to be more difficult to implement for two reasons.
首先，建立一个世界模型很难——世界是复杂的，我们获得的信息是嘈杂和不完整的。这是 LeCun 所称的缺失的世界模型，大脑皮层以某种方式呈现了这一模型。没有世界模型，就不可能模拟行为并预测其后果。
First, building a model of the world is hard—the world is complex and the information we get about it is noisy and incomplete. This is LeCun’s missing world model that the neocortex somehow renders. Without a world model, it is impossible to simulate actions and predict their consequences.
基于模型的强化学习之所以困难的第二个原因是，选择模拟什么很难。在马文·明斯基将时间信用分配问题确定为人工智能的障碍的同一篇论文中，他还提出了所谓的“搜索问题”：在大多数现实世界中，不可能搜索所有可能的选项。以国际象棋为例。构建国际象棋游戏的世界模型相对简单（规则是确定性的，你知道所有棋子、所有动作以及棋盘上的所有方格）。但在国际象棋中，你无法搜索所有可能的未来动作；国际象棋中的可能性树的分支路径比宇宙中的原子还要多。因此，问题不仅在于构建外部世界的内部模型，还在于弄清楚如何探索它。
The second reason model-based reinforcement learning is hard is that choosing what to simulate is hard. In the same paper that Marvin Minsky identified the temporal credit assignment problem as an impediment to artificial intelligence, he also identified what he called the “search problem”: In most real-world situations, it is impossible to search through all possible options. Consider chess. Building a world model of the game of chess is relatively trivial (the rules are deterministic, you know all the pieces, all their moves, and all the squares of the board). But in chess, you cannot search through all the possible future moves; the tree of possibilities in chess has more branching paths than there are atoms in the universe. So the problem is not just constructing an inner model of the external world but also figuring out how to explore it.
但很显然，早期哺乳动物的大脑以某种方式解决了搜索问题。让我们看看它是如何解决的。
And yet clearly, somehow, the brains of early mammals solved the search problem. Let’s see how.
十三
13
基于模型的强化学习
Model-Based Reinforcement Learning
TD-Gammon成功后，人们尝试将 Sutton 的时间差分学习（一种无模型强化学习）应用于国际象棋等更复杂的棋盘游戏。结果令人失望。
AFTER THE SUCCESS of TD-Gammon, people tried to apply Sutton’s temporal difference learning (a type of model-free reinforcement learning) to more complex board games like chess. The results were disappointing.
虽然时间差分学习等无模型方法在西洋双陆棋和某些视频游戏中表现良好，但它们在更多像国际象棋这样的复杂游戏。问题在于，在复杂的情况下，无模型学习（不包含任何规划或对可能的未来进行演练）不擅长发现目前看起来不太好但对未来有好处的举动。
While model-free approaches like temporal difference learning can do well in backgammon and certain video games, they do not perform well in more complex games like chess. The problem is that in complex situations, model-free learning—which contains no planning or playing out of possible futures—is not good at finding the moves that don’t look great right now but set you up well for the future.
2017 年，谷歌 DeepMind 发布了一个名为 AlphaZero 的人工智能系统，该系统不仅在国际象棋比赛中取得了超越人类的表现，而且在围棋比赛中也取得了超越人类的表现，击败了世界围棋冠军李世石。围棋是一种古老的中国棋盘游戏，比国际象棋更复杂；国际象棋的棋盘位置可能性是国际象棋的万亿倍。围棋比象棋好。
In 2017, Google’s DeepMind released an AI system called AlphaZero that achieved superhuman performance in not only the game of chess but also the game of Go, beating the world Go champion Lee Sedol. Go is an ancient Chinese board game that is even more complex than chess; there are a trillion upon trillion times more possible board positions in Go than in chess.
AlphaZero 是如何在围棋和国际象棋中取得超人表现的？AlphaZero 是如何在时间差分学习无法实现的地方取得成功的？关键区别在于 AlphaZero模拟了未来的可能性。与 TD-Gammon 一样，AlphaZero 是一个强化学习系统——它的策略不是用专家规则编入的，而是通过反复试验来学习的。但与TD-Gammon 不同的是，AlphaZero 是一种基于模型的强化学习算法；AlphaZero 在决定下一步做什么之前，会搜索可能的未来动作。
How did AlphaZero achieve superhuman performance at Go and chess? How did AlphaZero succeed where temporal difference learning could not? The key difference was that AlphaZero simulated future possibilities. Like TD-Gammon, AlphaZero was a reinforcement learning system—its strategies were not programmed into it with expert rules but learned through trial and error. But unlike TD-Gammon, AlphaZero was a model-based reinforcement learning algorithm; AlphaZero searched through possible future moves before deciding what to do next.
图片来自https://en.wikipedia.org/wiki/Go_(game)#/media/File:FloorGoban.JPG
Picture from https://en.wikipedia.org/wiki/Go_(game)#/media/File:FloorGoban.JPG
对手走完棋后，AlphaZero 会暂停，选择要考虑的棋步，然后进行数千次模拟，看看在选定的棋步下，整个游戏会如何进行。在进行一系列模拟后，AlphaZero 可能会发现，当它走 A 步时，它在 40 场想象比赛中赢得了 35 场；当它走 B 步时，它在 40 场想象比赛中赢得了 39 场；对于许多其他可能的下一步棋，AlphaZero 也是如此。然后，AlphaZero 可以选择它在想象比赛中获胜率最高的棋步。
After its opponent moved, AlphaZero would pause, select moves to consider, and then play out thousands of simulations of how the entire game might go given those selected moves. After running a set of simulations, AlphaZero might see that it won thirty-five out of the forty imagined games when it made move A, thirty-nine of the forty imagined games when it made move B, and so on for many other possible next moves. AlphaZero could then pick the move where it had won the highest ratio of imagined games.
当然，这样做会带来搜索问题；即使有了谷歌的超级计算机，也需要超过一百万年的时间才能模拟出任意一个棋子未来可能采取的所有行动。围棋中的棋盘位置。然而 AlphaZero 在半秒内就完成了这些模拟。怎么做到的？它并没有模拟数万亿种可能的未来；它只模拟了一千种未来。换句话说，它优先考虑了。
Doing this, of course, comes with the search problem; even armed with Google’s supercomputers, it would take well over a million years to simulate every possible future move from an arbitrary board position in Go. And yet AlphaZero ran these simulations within half a second. How? It didn’t simulate the trillions of possible futures; it simulated only a thousand futures. In other words, it prioritized.
有许多算法可以决定如何在庞大的可能性树中优先搜索哪些分支。谷歌地图在搜索从 A 点到 B 点的最佳路线时使用这种算法。但 AlphaZero 使用的搜索策略不同，它为了解真实大脑的运作方式提供了独特的见解。
There are many algorithms for deciding how to prioritize which branches to search through in a large tree of possibilities. Google Maps uses such an algorithm when it searches for the optimal route from point A to point B. But the search strategy used by AlphaZero was different and offered unique insight into how real brains might work.
我们已经讨论了在时间差异学习中，参与者如何根据对棋盘位置的直觉来预测下一步的最佳行动，而无需任何计划。AlphaZero 只是简单地扩展了这个架构中，AlphaZero 选择了多个其参与者认为最好的步骤，而不是选择单个参与者认为最好的步骤。AlphaZero 并没有假设参与者是正确的（但事实并非总是如此），而是使用搜索来验证参与者的猜测。AlphaZero 实际上是在对参与者说：“好吧，如果你认为步骤 A 是最好的步骤，那么让我们看看如果我们真的移动了 A，游戏会如何进行。”然后 AlphaZero 还探索了参与者的其他猜测，考虑了参与者建议的第二和第三佳步骤（对参与者说：“好吧，但是如果你不采取步骤 A，你的下一个最佳猜测是什么？也许步骤 B 的结果会比你想象的更好”）。
We already discussed how in temporal difference learning an actor learns to predict the best next move based on a hunch about the board position, doing so without any planning. AlphaZero simply expanded on this architecture. Instead of picking the single move its actor believed was the best next move, it picked multiple top moves that its actor believed were the best. Instead of just assuming its actor was correct (which it would not always be), AlphaZero used search to verify the actor’s hunches. AlphaZero was effectively saying to the actor, “Okay, if you think move A is the best move, let’s see how the game would play out if we did move A.” And AlphaZero then also explored other hunches of the actor, considering the second and third best moves the actor was suggesting (saying to the actor, “Okay, but if you didn’t take move A, what would your next best hunch be? Maybe move B will turn out even better than you think”).
它的优雅之处在于，从某种意义上说，AlphaZero 只是对 Sutton 时间差分学习的巧妙阐述，而不是重新发明。它使用搜索不是从逻辑上考虑所有未来的可能性（在大多数情况下这是不可能的），而是简单地验证和扩展演员-评论家系统已经产生的猜测。我们将看到，这种方法在原则上可能与哺乳动物解决搜索问题的方式相似。
What is elegant about this is that AlphaZero was, in some sense, just a clever elaboration on Sutton’s temporal difference learning, not a reinvention of it. It used search not to logically consider all future possibilities (something that is impossible in most situations) but to simply verify and expand on the hunches that an actor-critic system was already producing. We will see that this approach, in principle, may have parallels to how mammals navigate the search problem.
虽然围棋是最复杂的棋盘游戏之一，但它仍然比在现实世界中移动时模拟未来的任务简单得多。首先，围棋中的动作是离散的（从给定的棋盘位置来看，只有大约两百个可能的后续动作），而在现实世界中，动作是连续的（有无数种可能的围棋的三大特点是：一是围棋的规则是确定的、完整的，而现实世界的信息是嘈杂的、不完整的。二是围棋的奖励很简单（要么赢要么输），而现实世界中的动物有竞争性的需求，而且这些需求会随着时间而变化。因此，虽然 AlphaZero 是一个巨大的飞跃，但人工智能系统还远远不能在具有连续动作空间、不完整的世界信息和复杂奖励的环境中执行规划。
While Go is one of the most complex board games, it is still far simpler than the task of simulating futures when moving around in the real world. First, the actions in Go are discrete (from a given board position, there are only about two hundred possible subsequent next moves), whereas in the real world actions are continuous (there are an infinite number of possible body and navigational paths). Second, the information about the world in Go is deterministic and complete, whereas in the real world it is noisy and incomplete. And third, the rewards in Go are simple (you either win or lose the game), but in the real world, animals have competing needs that change over time. And so, while AlphaZero was a huge leap forward, AI systems are still far from performing planning in environments with a continuous space of actions, incomplete information about the world, and complex rewards.
然而，与 AlphaZero 等现代人工智能系统相比，哺乳动物大脑规划的最关键优势不在于它们能够利用连续的行动空间、不完整的信息或复杂的奖励进行规划，而在于哺乳动物大脑能够根据情况灵活地改变其规划方法。AlphaZero——这种策略只适用于棋盘游戏——每一步都采用相同的搜索策略。然而在现实世界中，不同的情况需要不同的策略。哺乳动物大脑模拟的出色之处不太可能是某种尚未发现的特殊搜索算法；更可能是哺乳动物大脑采用不同策略的灵活性。有时我们会停下来模拟我们的选择，但有时我们根本不模拟事情，只是本能地采取行动（大脑以某种方式智能地决定何时做每件事）。有时我们会停下来考虑可能的未来，但有时我们会停下来模拟一些过去的事件或其他过去的选择（大脑以某种方式选择何时做每件事）。有时我们会在计划中想象丰富的细节，执行每个详细的子任务，有时我们只呈现计划的总体思路（大脑以某种方式智能地选择模拟的正确粒度）。我们的大脑是如何做到这一点的？
However, the most critical advantage of planning in mammalian brains over modern AI systems like AlphaZero is not their ability to plan with continuous action spaces, incomplete information, or complex rewards, but instead simply the mammalian brain’s ability to flexibly change its approach to planning depending on the situation. AlphaZero—which applied only to board games—employed the same search strategy with every move. In the real world, however, different situations call for different strategies. The brilliance of simulation in mammal brains is unlikely to be some special yet-to-be-discovered search algorithm; it is more likely to be the flexibility with which mammal brains employ different strategies. Sometimes we pause to simulate our options, but sometimes we don’t simulate things at all and just act instinctually (somehow brains intelligently decide when to do each). Sometimes we pause to consider possible futures, but other times we pause to simulate some past event or alternative past choices (somehow brains select when to do each). Sometimes we imagine rich details in our plans, playing out each individual detailed subtask, and sometimes we render just the general idea of the plan (somehow brains intelligently select the right granularity of our simulation). How do our brains do this?
前额皮质与控制内部模拟
Prefrontal Cortex and Controlling the Inner Simulation
20 世纪 80 年代，一位名叫安东尼奥·达马西奥的神经学家探望了一位中风的病人，简称“L”。L 躺在床上，双眼睁开，面无表情。她一动不动，说不出话，但并没有瘫痪。她有时会用完美的动作灵巧地掀起毯子盖住自己；她会看向一个移动的物体，当有人叫她的名字时，她能清楚地识别出来。但她什么也没做，什么也没说。当人们看着她的眼睛时，他们说似乎“她在那儿，但又不在那儿”。
In the 1980s, a neuroscientist named Antonio Damasio visited one of his patients—referred to as “L”—who had suffered a stroke. L lay in bed with her eyes open and a blank expression on her face. She was motionless and speechless, but she wasn’t paralyzed. She would, at times, lift up the blanket to cover herself with perfectly fine motor dexterity; she would look over at a moving object, and she could clearly recognize when someone spoke her name. But she did and said nothing. When looking into her eyes, people said it seemed that “she was there but not there.”
视觉、躯体感觉或听觉新皮质受损的中风患者会遭受感知障碍（例如失明或失聪）。但 L 没有表现出这些症状；她的中风发生在前额叶新皮质的特定区域。L 患上了运动不能性缄默症——这是一种悲惨而怪异的疾病，由前额叶皮质的某些区域受损引起，患者能够正常移动并理解事物，但他们不会移动、说话或关心关于任何事情。
Stroke victims with damage to the visual, somatosensory, or auditory neocortex suffer from impairments to perception (such as blindness or deafness). But L showed none of these symptoms; her stroke occurred in a specific region of her prefrontal neocortex. L had developed akinetic mutism—a tragic and bizarre condition caused by damage to certain regions of the prefrontal cortex in which people are able to move and understand things just fine but they don’t move, speak, or care about anything at all.
六个月后，和许多中风患者一样，L 开始康复，因为她大脑皮层的其他区域重新映射，以弥补受损区域。当 L 慢慢开始说话时，达马西奥问她过去六个月的经历。虽然 L 对此记忆不多，但她确实记得开始说话前几天的情况。她将这次经历描述为无法说话，因为她无话可说。她声称她的大脑完全“空虚”，什么都不“重要”。她声称她完全能够跟上周围的谈话，但她“感觉没有‘意愿’去回复。”看来L已经失去回复的意思了。
After six months, as with many stroke patients, L began to recover as other areas of her neocortex remapped themselves to compensate for the damaged area. As L slowly began to speak again, Damasio asked her about her experience over the prior six months. Although L had little memory of it, she did recall the few days before beginning to speak. She described the experience as not talking because she had nothing to say. She claimed her mind was entirely “empty” and that nothing “mattered.” She claimed that she was fully able to follow the conversations around her, but she “felt no ‘will’ to reply.” It seems that L had lost all intention.
图 13.2
Figure 13.2
原创艺术由 Mesa Schumacher 创作
Original art by Mesa Schumacher
所有哺乳动物的大脑皮层都可以分为两半。后半部分是感觉大脑皮层，包含视觉、听觉和躯体感觉区域。我们在第 11 章中回顾的关于大脑皮层的所有内容都是关于感觉大脑皮层的——它是对外部世界的模拟呈现的地方，要么将其模拟与传入的感觉数据相匹配（通过推理感知），要么通过模拟替代现实（想象）。但感觉大脑皮层只是大脑如何运作的谜题的一半。大脑皮层是如何运作的。第一批哺乳动物的大脑皮层，就像现代老鼠和人类的大脑皮层一样，在大脑前半部分还有另一个组成部分：额叶大脑皮层。
The neocortex of all mammals can be separated into two halves. The back half is the sensory neocortex, containing visual, auditory, and somatosensory areas. Everything about the neocortex we reviewed in chapter 11 was about sensory neocortex—it is where a simulation of the external world is rendered, either matching its simulation to incoming sensory data (perception by inference) or by simulating alternative realities (imagination). But the sensory neocortex was only half the puzzle of how the neocortex works. The neocortex of the first mammals, just like that of modern rats and humans, had another component, found in the front half: the frontal neocortex.
人类大脑的前额皮质包含三个主要亚区域：运动皮质、颗粒前额皮质 (gPFC) 和无颗粒前额皮质 (aPFC)。颗粒和无颗粒这两个词根据颗粒细胞的存在来区分前额皮质的各个部分，颗粒细胞位于皮质的第四层。在颗粒前额皮质中，皮质包含正常的六层神经元。然而，在无颗粒前额皮质中，皮质的第四层（颗粒细胞所在的地方）却奇怪地缺失了。*因此，前额皮质中缺少第四层的部分称为无颗粒前额皮质 (aPFC)，而前额皮质中包含第四层的部分称为颗粒前额皮质 (gPFC)。目前还不清楚为什么额叶皮层的某些区域缺少一整层大脑皮层，但我们将在接下来的章节中探讨一些可能性。
The frontal neocortex of a human brain contains three main subregions: motor cortex, granular prefrontal cortex (gPFC), and agranular prefrontal cortex (aPFC). The words granular and agranular differentiate parts of the prefrontal cortex based on the presence of granule cells, which are found in layer four of the neocortex. In the granular prefrontal cortex, the neocortex contains the normal six layers of neurons. However, in the agranular prefrontal cortex, the fourth layer of neocortex (where granule cells are found) is weirdly missing.* Thus, the parts of the prefrontal cortex that are missing layer four are called the agranular prefrontal cortex (aPFC), and the parts of the prefrontal cortex that contain a layer four are called the granular prefrontal cortex (gPFC). It is still unknown why, exactly, some areas of frontal cortex are missing an entire layer of neocortex, but we will explore some possibilities in the coming chapters.
颗粒状前额皮质在早期灵长类动物中进化得晚得多，我们将在突破#4中了解它的全部内容。运动皮质在第一批哺乳动物之后、第一批灵长类动物之前进化（我们将在下一章中了解运动皮质）。但是，无颗粒状前额皮质 (aPFC) 是最古老的额叶区域，在第一批哺乳动物中进化而来。达马西奥的病人 L 的 aPFC 受损了。aPFC 是如此古老，对大脑皮层的正常运作至关重要，以至于当它受损时，L 就失去了作为人类的核心意义——或者更具体地说，失去了作为人类的意义。成为哺乳动物。
The granular prefrontal cortex evolved much later in early primates, and we will learn all about it in breakthrough #4. The motor cortex evolved after the first mammals but before the first primates (we will learn about the motor cortex in the next chapter). But the agranular prefrontal cortex (aPFC) is the most ancient of frontal regions and evolved in the first mammals. It was the aPFC that was damaged in Damasio’s patient L. The aPFC is so ancient and fundamental to the proper functioning of the neocortex that when damaged, L was deprived of something central to what it means to be human—or, more specifically, what it means to be a mammal.
在第一批哺乳动物中，整个额叶皮层只是无颗粒前额叶皮层。所有现代哺乳动物都含有从第一批哺乳动物遗传而来的无颗粒前额叶皮层。要了解 L 的运动不能性缄默症以及哺乳动物如何决定何时以及模拟什么，我们必须首先回溯进化时钟，探索 aPFC 在第一批哺乳动物大脑中的功能。
In the first mammals, the entire frontal cortex was just agranular prefrontal cortex. All modern mammals contain an agranular prefrontal cortex inherited from the first mammals. To understand L’s akinetic mutism and how mammals decide when and what to simulate, we must first roll back the evolutionary clock to explore the function of the aPFC in the brains of the first mammals.
似乎在早期哺乳动物中，感觉新皮层是模拟被呈现，而额叶新皮质是控制模拟的地方——额叶新皮质决定何时想象什么。额叶新皮质受损的老鼠失去了触发模拟的能力；它不再参与替代性尝试和错误，情景记忆回忆，或反事实学习。这在许多方面损害了老鼠的能力。它们在解决需要预先规划的空间导航挑战时变得更糟，例如当它们被放置在迷宫中全新的起始位置时。老鼠会做出更懒惰的选择，通常会选择更简单的路径，即使这些路径提供的奖励要少得多，就好像老鼠无法停下来尝试每个选项，并意识到努力是值得的。如果没有情景记忆，他们就无法在内心回忆起危险线索的旧记忆，因此更有可能重复过去的错误。
It seems that in early mammals, the sensory neocortex was where simulations were rendered, and the frontal neocortex was where simulations were controlled—it is the frontal neocortex that decided when and what to imagine. A rat with a damaged frontal neocortex loses the ability to trigger simulations; it no longer engages in vicarious trial and error, episodic-memory recall, or counterfactual learning. This impairs rats in many ways. They become worse at solving spatial-navigation challenges that require preplanning, such as when they are placed in completely new starting locations in a maze. Rats make lazier choices, often taking the easier paths even if they offer substantially fewer rewards, as if the rats are unable to stop and play out each option and realize that the effort is worth it. And without episodic memory, they fail to internally recall old memories of dangerous cues and are thereby more likely to repeat past mistakes.
图 13.3：第一批哺乳动物和现代人类的额叶区域
Figure 13.3: The frontal regions of the first mammals and of modern humans
原创艺术由 Mesa Schumacher 创作
Original art by Mesa Schumacher
即使那些只有部分额叶损伤的老鼠仍保留着触发这些模拟的能力，但它们仍然很难监控这些模拟的“计划”的展开。患有前额叶皮质损伤的老鼠很难记住它们在正在进行的计划中的位置，不按顺序采取行动，并且不必要地重复已经完成的动作。aPFC 受损的大鼠也会变得冲动；它们会在需要等待和等待的任务中过早做出反应。耐心去取食物。
Even those rats with only partial frontal damage who retain some ability to trigger these simulations still struggle to monitor these simulated “plans” as they unfold. Rats with aPFC damage struggle to remember where they are in an ongoing plan, do actions out of sequence, and unnecessarily repeat already completed actions. Rats with aPFC damage also become impulsive; they will prematurely respond in tasks that require them to wait and be patient to get food.
虽然额叶皮层和感觉皮层似乎发挥着不同的功能（额叶皮层触发模拟，感觉皮层呈现模拟），但它们都是皮层的不同区域，因此应该执行相同的基本计算。这带来了一个难题：额叶皮层，仅仅是皮层的另一个区域，是如何做一些看似与感觉皮层不同的事情的？为什么现代人 aPFC 受损后会失去意图？aPFC 如何在感觉皮层中触发模拟？它如何决定何时模拟某事？它如何决定模拟什么？
While the frontal and sensory cortices seem to serve different functions (the frontal neocortex triggers simulations, the sensory neocortex renders simulations), they are both different areas of the neocortex and thus should execute the same fundamental computation. This presents a conundrum: How does the frontal neocortex, simply another area of neocortex, do something so seemingly different from the sensory neocortex? Why does a modern human with a damaged aPFC become devoid of intention? How does the aPFC trigger simulations in the sensory neocortex? How does it decide when to simulate something? How does it decide what to simulate?
预测自己
Predicting Oneself
在感觉皮层的柱状结构中，主要输入来自外部传感器，例如眼睛、耳朵和皮肤。然而，无颗粒前额叶皮层的主要输入来自海马体、下丘脑和杏仁核。这表明 aPFC 处理位置序列、价态激活和内部情感状态的方式与感觉新皮层处理感觉信息序列的方式相同。那么，也许 aPFC 试图解释和预测动物自身行为的方式与感觉新皮层试图解释和预测外部感觉信息流的方式相同？
In a column of the sensory cortex, the primary input comes from external sensors, such as the eyes, ears, and skin. The primary input to the agranular prefrontal cortex, however, comes from the hippocampus, hypothalamus, and amygdala. This suggests that the aPFC treats sequences of places, valence activations, and internal affective states the way the sensory neocortex treats sequences of sensory information. Perhaps, then, the aPFC tries to explain and predict an animal’s own behavior the same way that the sensory neocortex tries to explain and predict the flow of external sensory information?
也许 aPFC 总是在观察老鼠的基底神经节驱动选择，并问自己：“为什么基底神经节会选择这个？” 例如，特定老鼠的 aPFC 可能会因此了解到，当老鼠醒来并出现这些特定的下丘脑激活时，它总是跑到河边喝水。然后 aPFC 可能会了解到这种行为的原因是“为了喝水”。然后在类似情况下，aPFC 可以在基底神经节触发任何行为之前预测动物会做什么——它可以预测当口渴时，动物会跑向附近的水源。换句话说，aPFC 学会模拟动物本身，推断它观察到的行为的意图，并利用这种意图预测动物下一步会做什么。
Perhaps the aPFC is always observing a rat’s basal-ganglia-driven choices and asking, “Why did the basal ganglia choose this?” The aPFC of a given rat might thereby learn, for example, that when the rat wakes up and has these specific hypothalamic activations, it always runs down to the river and consumes water. The aPFC might then learn that the why of such behavior is “to get water.” Then in similar circumstances, the aPFC can predict what the animal will do before the basal ganglia triggers any behavior—it can predict that when thirsty, the animal will run toward nearby water. In other words, the aPFC learns to model the animal itself, inferring the intent of behavior it observes, and uses this intent to predict what the animal will do next.
虽然意图在哲学上听起来有些模糊，但它在概念上与感觉皮层构建感觉信息解释的方式并无不同。当你看到一个暗示三角形的视觉错觉时（甚至当你在一个物体上看到一个三角形（当没有三角形时），你的感觉皮层就会构建一个解释，这就是你所感知到的三角形。这个解释——三角形——不是真实的，而是构建的。这是感觉皮层用来进行预测的一种计算技巧。对三角形的解释使你的感觉皮层能够预测如果你伸手去抓它、打开灯或试图从另一个角度看它，会发生什么。
As philosophically fuzzy as intent might sound, it is conceptually no different from how the sensory cortex constructs explanations of sensory information. When you see a visual illusion that suggests a triangle (even when there is no triangle), your sensory neocortex constructs an explanation of it, which is what you perceive—a triangle. This explanation—the triangle—is not real; it is constructed. It is a computational trick that the sensory neocortex uses to make predictions. The explanation of the triangle enables your sensory cortex to predict what would happen if you reached out to grab it or turned the light on or tried to look at it from another angle.
第一批哺乳动物的额叶新皮质与感觉新皮质
Frontal vs Sensory Neocortex in the First Mammals
额叶新皮质 FRONTAL NEOCORTEX
|
感觉新皮层 SENSORY NEOCORTEX
|
自我模型 A self model |
世界典范 A world model |
从海马体、杏仁核和下丘脑获取输入 Gets input from hippocampus, amygala, and hypothalamus
|
从感觉器官获取输入 Gets input from sensory organs
|
“我这么做是因为我想喝水” “I did this because I want to get to water” |
“我看到这个是因为那里有一个三角形” “I see this because there is a triangle right there” |
试图预测动物下一步会做什么 Tries to predict what the animal will do next |
尝试预测外部物体下一步会做什么 Tries to predict what external objects will do next |
记录研究证实了 aPFC 为动物的目标创建模型的想法。如果你查看大鼠 aPFC 的记录，你可以看到编码大鼠正在执行的任务的活动模式——特定神经元群仅在复杂任务序列中的特定位置选择性地激发，可靠地跟踪进度朝着想象中的目标。
Recording studies corroborate the idea that the aPFC creates a model of an animal’s goals. If you look at a recording of the aPFC of a rat, you can see patterns of activity that encode the task a rat is performing—with specific populations of neurons selectively firing only at specific locations within a complex task sequence, reliably tracking progress toward an imagined goal.
额叶皮质中的这种自我模型在进化上有什么用处？为什么要通过构建“意图”来“解释”自己的行为？事实证明，这可能是哺乳动物选择何时模拟事物以及如何选择模拟什么的方式。解释自己的行为可能会解决搜索问题。让我们看看如何解决。
What is the evolutionary usefulness of this model of self in the frontal cortex? Why try to “explain” one’s own behavior by constructing “intent”? It turns out, this might be how mammals choose when to simulate things and how to select what to simulate. Explaining one’s own behavior might solve the search problem. Let’s see how.
哺乳动物如何做出选择
How Mammals Make Choices
让我们以一只老鼠为例，它在迷宫中穿行，当它到达岔路口时，它会选择往哪个方向走。往左边走，就会到达向右走可以找到水，向右走可以找到食物。在这些情况下，就会发生替代性试错，它分为三个步骤。
Let’s take the example of a rat navigating a maze and making a choice as to which direction to go when it reaches a fork. Going to the left leads to water, going to the right leads to food. It is these situations when vicarious trial and error occurs, and it occurs in three steps.
图 13.4：关于哺乳动物如何做出深思熟虑的选择的推测
Figure 13.4: Speculations on how mammals make deliberative choices
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
步骤#1：触发模拟
Step #1: Triggering Simulation
aPFC 中的柱状图可能始终处于以下三种状态之一：(1) 沉默；它观察到的行为没有识别任何特定意图（例如视觉皮层中的柱状图没有识别图像中任何有意义的内容）；(2) 额叶皮层中的许多或所有柱状图识别出意图并预测相同的下一步行为（“哦！我们显然要从这里向左走”）；或 (3) 额叶皮层中的许多柱状图识别出意图但预测出不同且不一致的行为（一些柱状图预测“我要从左边走去喝水！”而其他柱状图预测“我要从右边走去吃东西！”）。在最后一种状态下，aPFC 中的柱状图预测不一致，这时 aPFC 可能最有用。事实上，当出现问题或发生意外情况时，哺乳动物的 aPFC 会最兴奋在一项正在进行的任务中。
The columns in the aPFC might always be in one of three states: (1) silent; the behavior it observes is not recognizing any specific intent (like a column in the visual cortex not recognizing anything meaningful in an image); (2) many or all columns in the frontal cortex recognize an intent and predict the same next behavior (“Oh! We are obviously about to go to the left here”); or (3) many columns in the frontal cortex recognize an intent but predict different and inconsistent behaviors (some columns predict “I’m going to go down to the left to get water!” and other columns predict “I’m going to go to the right to get food!”). It might be in this last state, in which aPFC columns don’t agree in their predictions, where the aPFC might be most useful. Indeed, the aPFC of mammals gets most excited when something goes wrong or something unexpected happens in an ongoing task.
预测不一致的程度是衡量不确定性的标准。原则上，这就是许多最先进的机器学习模型衡量不确定性的方法：一组不同的模型做出预测，预测越不一致，不确定性就越大据报道。
The degree of disagreement of predictions is a measure of uncertainty. This is, in principle, how many state-of-the-art machine learning models measure uncertainty: An ensemble of different models makes predictions, and the more divergent such predictions, the more uncertainty there is reported to be.
可能正是这种不确定性引发了模拟。aPFC 可以通过直接连接到特定部分来触发全局暂停信号基底神经节和 aPFC 激活已被证明与存在不确定性。正如我们在上一章中看到的，当事情发生变化或变得困难（即不确定）时，动物会停下来进行间接试错。也有可能在基底神经节中测量不确定性；也许存在平行的行动者-评论家系统，每个系统都对下一个最佳行动做出独立的预测，而它们预测的分歧就是触发暂停的原因。
It might be this uncertainty that triggers simulations. The aPFC can trigger a global pause signal through connecting directly to specific parts of the basal ganglia, and aPFC activation has been shown to correlate with levels of uncertainty. And as we saw in the last chapter, it is exactly when things change or are hard (i.e., are uncertain) that animals pause to engage in vicarious trial and error. It is also possible that uncertainty could be measured in the basal ganglia; perhaps there are parallel actor-critic systems, each making independent predictions about the next best action, and the divergence in their predictions is what triggers pausing.
无论如何，这为哺乳动物的大脑如何应对决定何时进行模拟事物的努力提供了一个推测。如果事件按照预期展开，就没有理由浪费时间和精力模拟选项，让基底神经节驱动决策（无模型学习）更容易，但当出现不确定性时（出现新事物、某些偶然性被打破或成本接近收益），就会触发模拟。
Either way, this provides a speculation for how mammal brains tackle the challenge of deciding when to go through the effort of simulating things. If events are unfolding as one would expect, there is no reason to waste time and energy simulating options, and it is easier just to let the basal ganglia drive decisions (model-free learning), but when uncertainty emerges (something new appears, some contingency is broken, or costs are close to the benefits), then simulation is triggered.
步骤2：模拟选项
Step #2: Simulating Options
好吧，老鼠停顿了一下，决定用模拟来解决它的不确定性——现在怎么办？这又把我们带回到搜索问题。迷宫中的老鼠可以做上亿件事中的任何一件事，那么它如何决定要模拟什么呢？
Okay, so the rat paused and decided to use simulation to resolve its uncertainty—now what? This brings us back to the search problem. A rat in a maze could do any one of a billion things, so how does it decide what to simulate?
我们看到了 AlphaZero 如何解决这个问题：它执行了它已经预测的最佳动作。这个想法与已知的新皮质柱和基底神经节非常吻合。aPFC 不会坐在那里梳理所有可能的动作，而是专门探索它已经预测动物将采取的路径。换句话说，aPFC 会搜索 aPFC 不同柱已经预测的特定选项。一组柱预测一直向左走到水边，另一组柱预测向右走，因此只需运行两个不同的模拟。
We saw how AlphaZero solved this problem: It played out the top moves it was already predicting were the best. This idea aligns quite nicely with what is known about neocortical columns and the basal ganglia. The aPFC doesn’t sit there combing through every possible action, instead it specifically explores the paths that it is already predicting an animal will take. In other words, the aPFC searches through the specific options that different columns of the aPFC are already predicting. One set of columns predicted going left all the way to water, and another predicted going right, so there are only two different simulations to run.
动物暂停后，aPFC 中的不同列会轮流预测动物会做什么。一组列预测动物向左走，并沿着这条路一直走到水边。另一组列预测动物向右走，并沿着这条路一直走到食物边。
After an animal pauses, different columns in the aPFC take turns playing out their predictions of what they think the animal will do. One group of columns plays out going left and following that path all the way to water. Another group of columns plays out going right and following that path all the way to get food.
aPFC 和感觉新皮质之间的连接很有启发性；aPFC 广泛投射到感觉的弥散区域皮质，并已被证明能显著调节感觉新皮层的活动。特别是当老鼠进行这种替代性试错行为时，aPFC 和感觉大脑皮层变得独特地同步。一种推测是，aPFC 触发感觉新皮层来呈现对世界的特定模拟。aPFC 首先问：“如果我们向左走会发生什么？”然后感觉新皮层呈现左转的模拟，然后将其传回 aPFC。然后 aPFC 说：“好的，那么如果我们继续前进会发生什么？”感觉新皮层再次呈现该模拟，依此类推，直到 aPFC 中建模的想象目标。
The connectivity between the aPFC and the sensory neocortex is revealing; the aPFC projects extensively to diffuse regions of the sensory cortex and has been shown to dramatically modulate the activity of the sensory neocortex. And specifically when rats engage in this vicarious trial and error behavior, the activity in the aPFC and the sensory cortex become uniquely synchronized. One speculation is that the aPFC is triggering the sensory neocortex to render a specific simulation of the world. The aPFC first asks, “What happens if we go to the left?” The sensory neocortex then renders a simulation of turning left, which then passes back to the aPFC. The aPFC then says, “Okay, and then what happens if we keep going forward?” which the sensory neocortex renders again, and so on and so forth all the way to the imagined goal modeled in the aPFC.
或者，可能是基底神经节决定了这些模拟过程中采取的行动。这将更接近 AlphaZero 的工作方式——它根据其无模型参与者预测的最佳行动来选择模拟行动。在这种情况下，aPFC 会选择模拟基底神经节的哪些不同行动预测，但基底神经节将继续决定它想要在感觉新皮层呈现的想象世界中采取哪些行动。
Alternatively, it could be the basal ganglia that determines the actions taken during these simulations. This would be even closer to how AlphaZero worked—it selected simulated actions based on the actions its model-free actor predicted were best. In this case, it would be the aPFC that selects which of the divergent action predictions of the basal ganglia to simulate, but the basal ganglia would continue to decide which actions it wants to take in the imagined world rendered by the sensory neocortex.
步骤#3：选择一个选项
Step #3: Choosing an Option
大脑皮层模拟一系列动作，但是什么决定了老鼠实际上会朝哪个方向走呢？老鼠如何选择？这是一种推测。基底神经节已经有一个做出选择的系统。即使是古代脊椎动物在面对竞争刺激时也必须做出选择。基底神经节积累了竞争选择的投票，代表每个竞争动作的不同神经元群兴奋地不断增加，直到它通过选择阈值，此时已选择一项操作。
The neocortex simulates sequences of actions, but what makes the final decision as to what direction the rat will actually go? How does a rat choose? Here’s one speculation. The basal ganglia already has a system for making choices. Even the ancient vertebrates had to make choices when presented with competing stimuli. The basal ganglia accumulates votes for competing choices, with different populations of neurons representing each competing action ramping up in excitement until it passes a choice threshold, at which point an action is selected.
因此，随着替代性试错过程的展开，这些替代性行为重演的结果会在基底神经节中积累每个选择的投票——如果试错不是替代性的而是真实的，情况也是如此。如果基底神经节在想象喝水时比想象吃食物时更兴奋（以释放的多巴胺量来衡量），那么这些选择水的投票将很快超过选择阈值。基底神经节将接管行为，老鼠就会去取水。
And so, as the process of vicarious trial and error unfolds, the results of these vicarious replays of behavior accumulate votes for each choice in the basal ganglia—the same way it would if the trial and error were not vicarious but real. If the basal ganglia keeps getting more excited by imagining drinking water than by imagining eating food (as measured by the amount of dopamine released), then these votes for water will quickly pass the choice threshold. The basal ganglia will take over behavior, and the rat will go get water.
所有这些的突发效应是，aPFC间接地训练基底神经节，让它知道左转是更好的选择。基底神经节不知道感觉新皮质是在模拟当前世界还是想象的世界。基底神经节只知道，当它向左转时，它会得到强化。因此，当感觉新皮质回到迷宫开始时模拟当前世界时，基底神经节会立即尝试重复刚刚间接强化的行为。瞧——动物跑到左边去喝水。
The emergent effect of all this is that the aPFC vicariously trained the basal ganglia that left was the better option. The basal ganglia doesn’t know whether the sensory neocortex is simulating the current world or an imagined world. All the basal ganglia knows is that when it turned left, it got reinforced. Hence, when the sensory neocortex goes back to simulating the current world at the beginning of the maze, the basal ganglia immediately tries to repeat the behavior that was just vicariously reinforced. Voilà—the animal runs to the left to get water.
目标和习惯（或哺乳动物的内在二元性）
Goals and Habits (or the Inner Duality of Mammals)
20 世纪 80 年代初，剑桥大学的一位名叫托尼·迪金森 (Tony Dickinson) 的心理学家参与了当时流行的心理学实验：训练动物推动杠杆以获得奖励。迪金森提出了一个看似平凡的问题：如果在动物学会某种行为后降低其奖励的价值，会发生什么情况？假设你教一只老鼠，推动杠杆会从附近的装置中释放出食物颗粒。老鼠会快速地在推动杠杆和狼吞虎咽之间来回切换。现在假设有一天，完全脱离了带杠杆的盒子，你给老鼠同样的食物颗粒，并偷偷地在其中加入一种让老鼠感到恶心的化学物质。这会如何改变它们的行为？
In the early 1980s, a Cambridge psychologist by the name of Tony Dickinson was engaging in the popular psychology experiments of the time: training animals to push levers to get rewards. Dickinson was asking a seemingly mundane question: What happens if you devalue the reward of a behavior after the behavior is learned? Suppose you teach a rat that pushing a lever releases a food pellet from a nearby contraption. The rat will rapidly go back and forth between pushing the lever and gobbling up the food. Now suppose one day, completely out of the context of the box with the lever, you give the rat the same pellet and secretly lace it with a chemical that makes the rat feel sick. How does this change their behavior?
第一个结果是意料之中的：即使老鼠从短暂的恶心中恢复过来，它们也不再像以前那样觉得这些小球有食欲。当给老鼠一堆这样的小球时，它们吃得少得多。但更有趣的问题是：下次给老鼠看杠杆时会发生什么？如果动物只是受桑代克效应定律的支配，那么它们会跑到杠杆前，像以前一样快速地推它——杠杆已经得到过多次强化，而且还没有任何东西取消推动杠杆的行为。另一方面，如果动物确实能够模拟推动杠杆的后果，并意识到结果是它们不再喜欢的小球，它们就不会那么想推杠杆了。迪金森发现，经过这一程序后，与生病的老鼠相比，生病的老鼠推杠杆的次数减少了近 50%。那些没有的。
The first result is expected: Rats, even after they have long recovered from their short bout of nausea, no longer find the pellets as appetizing as they did before. When given a mound of such pellets, rats eat substantially fewer. But the more interesting question was this: What would happen the next time the rats were shown the lever? If animals are simply governed by Thorndike’s law of effect, then they would run up to the lever and push it as rapidly as before—the lever had been reinforced many times, and nothing had yet unreinforced the act of pushing the lever. On the other hand, if animals are indeed able to simulate the consequences of pushing the lever and realize that the result is a pellet that they no longer like, they won’t want to push the lever as much. What Dickinson found was that after this procedure, rats who had the pellet paired with illness pushed the lever almost 50 percent less than those that had not.
这与大脑皮层能够实现简单任务的观点一致哺乳动物（如老鼠）会间接模拟未来的选择，并根据想象的后果改变行为。但随着迪金森继续进行这些实验，他注意到一些奇怪的事情：在将药丸与疾病配对后，一些老鼠继续以同样大的力度，甚至更大的力度推杠杆。一些老鼠变得被他称为“对贬低不敏感”。他发现，这种差异仅仅是老鼠为了获得奖励而推杠杆的次数的结果。已经完成任务一百次的老鼠做了一件聪明的事——它们不再想再推一次杠杆食物贬值了。但是，已经完成这项任务 500 次的老鼠会跑到杠杆前，开始疯狂地推杠杆，即使食物贬值了。在所有这些测试中，从未给予食物颗粒；对贬值不敏感的那组老鼠只是继续推杠杆，从未得到任何奖励。
This is consistent with the idea that the neocortex enables even simple mammals such as rats to vicariously simulate future choices and change their behaviors based on the imagined consequences. But as Dickinson continued these experiments, he noticed something weird: Some rats continued to push the lever with as much, if not more, vigor after the pellet was paired with illness. Some rats became what he called “insensitive to devaluation.” The difference, he found, was merely a consequence of how many times the rats had pushed the lever to get a reward. Rats that had done the task one hundred times did the smart thing—they no longer wanted to push the lever once the food was devalued. But rats that had done the task five hundred times ran up to the lever and just started pushing it like crazy, even if the food was devalued. And in all these tests, the food pellets never were given; the group that had become insensitive to devaluation just kept pushing the lever without ever getting a reward.
迪金森发现了习惯。通过进行五百次这种行为，老鼠已经形成了一种自动运动反应，这种反应由感官提示触发，完全脱离了行为的更高层次目标。基底神经节接管了行为，而 aPFC 却没有停下来考虑这些行为会产生什么未来。这种行为重复了很多次，以至于 aPFC 和基底神经节没有检测到任何不确定性，因此动物没有停下来考虑后果。
Dickinson had discovered habits. By engaging in the behavior five hundred times, rats had developed an automated motor response that was triggered by a sensory cue and completely detached from the higher-level goal of the behavior. The basal ganglia took over behavior without the aPFC pausing to consider what future these actions would produce. The behavior had been repeated so many times that the aPFC and basal ganglia did not detect any uncertainty and therefore the animal did not pause to consider the consequences.
也许这是一种常见的体验。人们醒来后会看手机，却不会问自己为什么要看手机。他们不停地浏览 Instagram，即使有人问他们是否想继续浏览，他们会说“不”。当然，并非所有习惯都是坏的：你不会想着走路，但你走路却很完美；你不会想着打字，但想法却毫不费力地从你的大脑流向你的指尖；你不会想着说话，但想法却神奇地转化为舌头、嘴巴和喉咙的动作。
Perhaps this is a familiar experience. People wake up and look at their phones without asking themselves why they are choosing to look at their phones. They keep scrolling through Instagram even though if someone had asked them if they wanted to keep scrolling, they’d say “no.” Of course, not all habits are bad: You don’t think about walking, and yet you walk perfectly; you don’t think about typing, and yet the thoughts flow effortlessly from your mind to your fingertips; you don’t think about speaking, and yet thoughts magically convert themselves into a repertoire of tongue, mouth, and throat movements.
习惯是直接由刺激触发的自动化行为（它们不受模型影响）。它们是直接由基底神经节控制的行为。它们是哺乳动物大脑节省时间和精力的方式，避免不必要地进行模拟和规划。当这种自动化在正确的时间发生时，它使我们能够完成复杂的行为；当它发生在错误的时间时，我们就会做出错误的选择。
Habits are automated actions triggered by stimuli directly (they are model-free). They are behaviors controlled directly by the basal ganglia. They are the way mammalian brains save time and energy, avoiding unnecessarily engaging in simulation and planning. When such automation occurs at the right times, it enables us to complete complex behaviors easily; when it occurs at the wrong times, we make bad choices.
基于模型和无模型的决策方法之间的二元性在不同领域以不同的形式出现。在人工智能中，使用基于模型和无模型这两个术语。在动物心理学中，同样的二元性被描述为目标驱动行为和习惯性行为。而在行为经济学中，如丹尼尔·卡尼曼的著名著作《思考，快与慢》中所述，同样的二元性被描述为“系统 2”（慢思考）与“系统 1”（快思考）。在所有这些情况下，二元性都是相同的：人类，实际上所有哺乳动物（以及一些其他独立进化出模拟能力的动物）有时会停下来模拟他们的选择（基于模型、目标驱动、系统 2），有时会自动行动（无模型、习惯性、系统 1）。两者都没有更好，每种都有其优点和缺点。大脑试图智能地选择何时做每件事，但大脑并不总是能正确地做出这个决定，而这正是我们许多非理性行为的根源。
The duality between model-based and model-free decision-making methods shows up in different forms across different fields. In AI, the terms model-based and model-free are used. In animal psychology, this same duality is described as goal-driven behavior and habitual behavior. And in behavioral economics, as in Daniel Kahneman’s famous book Thinking, Fast and Slow, this same duality is described as “system 2” (thinking slow) versus “system 1” (thinking fast). In all these cases, the duality is the same: Humans and, indeed, all mammals (and some other animals that independently evolved simulation) sometimes pause to simulate their options (model-based, goal-driven, system 2) and sometimes act automatically (model-free, habitual, system 1). Neither is better; each has its benefits and costs. Brains attempt to intelligently select when to do each, but brains do not always make this decision correctly, and this is the origin of many of our irrational behaviors.
动物心理学中使用的语言很有启发性——一种行为是目标驱动的，另一种则不是。事实上，目标本身可能直到早期哺乳动物才进化出来。
The language used in animal psychology is revealing—one type of behavior is goal-driven and the other is not. Indeed, goals themselves may not have evolved until early mammals.
第一个进球的演变
The Evolution of the First Goal
正如对感官信息的解释不是真实的（即，您不会感知到您所看到的东西），意图也不是真实的；相反，它是一种计算技巧，用于预测动物下一步会做什么。
Just as the explanations of sensory information are not real (i.e., you don’t perceive what you see), so intent is not real; rather, it is a computational trick for making predictions about what an animal will do next.
这一点很重要：基底神经节没有意图或目标。像基底神经节这样的无模型强化学习系统是无意图的；它是一个简单地学习重复之前得到过强化的行为的系统。这并不是说这种无模型系统很愚蠢或没有动机；它们可以非常聪明，可以快速学习产生最大化奖励量的行为。但这些无模型系统没有“目标”，因为它们没有追求特定结果的意图。这就是为什么无模型强化学习系统很难解释的原因之一——当我们问“为什么人工智能系统会这样做？”时，我们实际上是在问这个问题其实没有答案。或者至少答案永远是一样的：因为它认为这是预期回报最多的选择。
This is important: The basal ganglia has no intent or goals. A model-free reinforcement learning system like the basal ganglia is intent-free; it is a system that simply learns to repeat behaviors that have previously been reinforced. This is not to say that such model-free systems are dumb or devoid of motivation; they can be incredibly intelligent and clever, and they can rapidly learn to produce behavior that maximizes the amount of reward. But these model-free systems do not have “goals” in the sense that they do not set out to pursue a specific outcome. This is one reason why model-free reinforcement learning systems are painfully hard to interpret—when we ask, “Why did the AI system do that?,” we are asking a question to which there is really no answer. Or at least, the answer will always be the same: because it thought that was the choice with the most predicted reward.
相比之下，aPFC确实有明确的目标——它想去冰箱吃草莓或去饮水机喝水。通过模拟以某种最终结果结束的未来，aPFC 有一个它试图实现的最终状态（目标）。这就是为什么至少在人们做出 aPFC 驱动（目标导向、基于模型、系统 2）选择的情况下，可以问一个人为什么做某事。
In contrast, the aPFC does have explicit goals—it wants to go to the fridge to eat strawberries or go to the water fountain to drink water. By simulating a future that terminates at some end result, the aPFC has an end state (a goal) that it seeks to achieve. This is why it is possible, at least in circumstances where people make aPFC-driven (goal-oriented, model-based, system 2) choices, to ask why a person did something.
感觉皮层中构建外部物体模型的大脑皮层微电路，可以重新用于构建目标，并改变额叶皮层中的行为以追求这些目标，这有点神奇。伦敦大学学院的卡尔·弗里斯顿（Karl Friston）是大脑皮层实现生成模型这一理念的先驱之一）将此称为“主动推理”。感觉皮层进行被动推理——仅仅解释和预测感觉输入。aPFC 进行主动推理——解释自己的行为，然后利用其预测来主动改变这种行为。通过暂停来演绎 aPFC 预测将要发生的事情，从而间接训练基底神经节，aPFC 正在重新利用新皮层生成模型进行预测以创造意志。
It is somewhat magical that the very same neocortical microcircuit that constructs a model of external objects in the sensory cortex can be repurposed to construct goals and modify behavior to pursue these goals in the frontal cortex. Karl Friston of University College London—one of the pioneers of the idea that the neocortex implements a generative model—calls this “active inference.” The sensory cortex engages in passive inference—merely explaining and predicting sensory input. The aPFC engages in active inference—explaining one’s own behavior and then using its predictions to actively change that behavior. By pausing to play out what the aPFC predicts will happen and thereby vicariously training the basal ganglia, the aPFC is repurposing the neocortical generative model for prediction to create volition.
当你暂停并模拟不同的晚餐选择，选择吃意大利面，然后开始漫长的行动序列去餐厅时，这是一个“自愿”的选择——你可以回答你为什么要上车；你知道你追求的最终状态。相比之下，当你只是出于习惯而行动时，你无法回答为什么你会这么做。
When you pause and simulate different dinner options, choose to get pasta, then begin the long action sequence to get to the restaurant, this is a “volitional” choice—you can answer why you are getting in the car; you know the end state you are pursuing. In contrast, when you act only from habit, you have no answer as to why you did what you did.
卡尔·弗里斯顿 (Karl Friston) 还对额叶皮质的某些部分缺少新皮质柱的第四层。第四层起什么作用？在感觉皮层中，第四层是原始感觉输入流入新皮质柱的地方。据推测，第四层的作用是推动新皮质柱的其余部分呈现与其传入感觉数据最匹配的模拟（推理感知）。有证据表明，当新皮质柱参与模拟时，第四层活动会下降，因为主动传入感觉输入受到抑制——这就是新皮质可以模拟当前没有经历过的事情（例如，想象一辆汽车主动推理表明，aPFC 构建意图，然后尝试预测与该意图一致的行为；换句话说，它试图使其意图成真。如果动物做了与 aPFC 构建的意图不一致的事情，aPFC 不会调整其意图模型以匹配行为，而是会调整行为：如果你口渴，而你的基底神经节决定朝没有水的方向走，aPFC 不会调整其意图模型以假设你不口渴，而是想暂停基底神经节的错误，并说服它转身走向水。因此，aPFC 花费很少的时间（如果有的话）尝试将其推断的意图与它看到的行为相匹配，因此它不需要大的甚至任何第四层。
Karl Friston also offers an explanation for the perplexing fact that some parts of the frontal cortex are missing the fourth layer of the neocortical column. What does layer four do? In the sensory cortex, layer four is where raw sensory input flows into a neocortical column. Layer four is speculated to have the role of pushing the rest of the neocortical column to render a simulation that best matches its incoming sensory data (perception by inference). There is evidence that when a neocortical column is engaged in simulating, layer-four activity declines as active incoming sensory input is suppressed—this is how the neocortex can render a simulation of something not currently experienced (e.g., imagining a car when looking at the sky). This is a clue. Active inference suggests that the aPFC constructs intent and then tries to predict behavior consistent with that intent; in other words, it tries to make its intent come true. If the animal does something inconsistent with the aPFC’s constructed intent, the aPFC doesn’t want to adjust its model of intent to match the behavior, it wants to adjust the behavior: if you are thirsty and your basal ganglia makes a decision to go in the direction that has no water, the aPFC doesn’t want to adjust its model of your intent to assume you are not thirsty, instead it wants to pause the basal ganglia’s mistake, and convince it to turn around and go toward the water. Thus, the aPFC spends very little, if any, time trying to match its inferred intent to the behavior it sees, and so it has no need for a large, or even any, layer four.
当然，aPFC 并非在进化过程中被设计用来理解动物的目标，而是通过首先模拟最初由基底神经节控制的行为来学习这些目标。aPFC 通过观察最初完全没有目标的行为来构建目标。只有在学习了这些目标之后，aPFC 才会开始控制行为：基底神经节最初是 aPFC 的老师，但随着哺乳动物的发育，这些角色互换，aPFC 成为基底神经节的老师。事实上，在大脑发育过程中，额叶皮层的无颗粒部分从第四层开始，然后在发育过程中逐渐萎缩并消失，留下第四层大部分是空的。也许这是构建自我模型的发展程序的一部分，首先将一个人的内部模型与其观察结果相匹配（因此从第 4 层开始），然后过渡到推动行为以匹配一个人的内部模型（因此不再需要第 4 层）。我们再次看到了进化中美丽的引导。
Of course, the aPFC isn’t evolutionarily programmed to understand the goals of the animal, instead it learns these goals by first modeling behavior originally controlled by the basal ganglia. The aPFC constructs goals by observing behavior that is originally entirely devoid of them. And only once these goals are learned does the aPFC begin to exert control over behavior: the basal ganglia begins as the teacher of the aPFC, but as a mammal develops, these roles flip, and the aPFC becomes the teacher of the basal ganglia. And indeed, during brain development, agranular parts of the frontal cortex begin with a layer four that then slowly atrophies and disappears during development, leaving layer four largely empty. Perhaps this is part of a developmental program for constructing a model of self, starting by matching one’s internal model to its observations (hence beginning with a layer 4), and then transitioning to pushing behavior to match one’s internal model (hence no need for a layer 4 anymore). Again we see a beautiful bootstrapping in evolution.
这也让我们对达马西奥的病人 L 的经历有了些许了解。这也解释了为什么她的脑袋是“空的”：她无法进行内在模拟。她没有任何想法。她没有意愿对任何事情做出回应，因为她的内在意图模型已经消失，而没有这个，她的大脑就无法设定哪怕是最简单的目标。而不幸的是，没有目标，一切都不重要。
This also offers some insight into the experience of Damasio’s patient L. It makes some sense why her head was “empty”: She was unable to render an inner simulation. She had no thoughts. She had no will to respond to anything because her inner model of intent was gone, and without that, her mind could not set even the simplest of goals. And without goals, tragically, nothing mattered.
哺乳动物如何控制自己：注意力、工作记忆和自我控制
How Mammals Control Themselves: Attention, Working Memory, and Self-Control
在典型的神经科学教科书中，额叶新皮质被归为四种功能：注意力、工作记忆、执行控制，以及我们已经看到的计划。这些功能之间的联系一直令人困惑；一个结构竟然可以发挥所有这些不同的作用，这似乎很奇怪。但从进化的角度来看，这些功能都密切相关，这是有道理的——它们都是控制新皮质模拟的不同应用。
In a typical neuroscience textbook, the four functions ascribed to the frontal neocortex are attention, working memory, executive control, and, as we have already seen, planning. The connecting theme of these functions has always been confusing; it seems odd that one structure would subserve all these distinct roles. But through the lens of evolution, it makes sense that these functions are all intimately related—they are all different applications of controlling the neocortical simulation.
记住一张模糊的照片鸭子还是兔子？当你在感知鸭子和兔子之间犹豫不决时，你的 aPFC 会来回推动你的视觉皮层在这两种解释之间来回移动。当你闭上眼睛时，你的 aPFC 可以触发对鸭子的内部模拟，当你睁开眼睛看着可能是鸭子或兔子的图片时，你的 aPFC 也可以使用相同的机制来触发对鸭子的内部模拟。在这两种情况下，aPFC 都在试图调用模拟；唯一的区别是，当你闭上眼睛时，模拟不受约束，而当你睁开眼睛时，模拟会受到约束，以与你所看到的内容保持一致。当 aPFC不受当前感官输入的约束时，它触发的模拟称为想象力，当它受到当前感官输入的约束时，它被称为注意力。但在这两种情况下，aPFC 原则上都在做同样的事情。
Remember the ambiguous picture of a duck or a rabbit? As you oscillate between perceiving a duck or a rabbit, it is your aPFC that is nudging your visual cortex back and forth between each interpretation. Your aPFC can trigger an internal simulation of ducks when you close your eyes, and your aPFC can use the same mechanism to trigger an internal simulation of ducks with your eyes open and you are looking at a picture that could be either a duck or a rabbit. In both cases, the aPFC is trying to invoke a simulation; the only difference is that with your eyes closed, the simulation is unconstrained, and with your eyes open it is constrained to be consistent with what you are seeing. The aPFC’s triggering of simulation is called imagination when it is unconstrained by current sensory input and attention when it is constrained by current sensory input. But in both cases, the aPFC is, in principle, doing the same thing.
注意的重点是什么？当老鼠在想象的模拟后选择一个动作序列时，它必须在沿着路径奔跑时坚持自己的计划。这比听起来要难。想象的模拟不会是完美的；老鼠不会预测它实际会经历的每种视觉、气味和环境轮廓。这意味着基底神经节经历的间接学习将与计划展开时的实际体验不同，因此，基底神经节可能无法正确实现预期的行为。
What is the point of attention? When a mouse selects an action sequence after its imagined simulation, it must stick to its plan as it runs down its path. This is harder than it sounds. The imagined simulation will not have been perfect; the mouse will not have predicted each sight, smell, and contour of the environment that it will actually experience. This means that the vicarious learning that the basal ganglia experienced will differ from the actual experience as the plan unfolds, and therefore, the basal ganglia may not correctly fulfill the intended behavior.
aPFC 解决这个问题的一种方法是利用注意力。假设一只老鼠的基底神经节通过反复试验学会了躲避鸭子并跑向兔子。在这种情况下，基底神经节在看到鸭子/兔子时会做出相反的反应，具体取决于发送给它的模式大脑皮层接收到了这种模糊的图像。如果前额叶皮质之前想象过看到一只兔子并跑向它，那么它就可以利用注意力来控制基底神经节的选择，以确保当老鼠看到这张模糊的图片时，它看到的是一只兔子，而不是一只鸭子。
One way the aPFC can solve this problem is using attention. Suppose a rat’s basal ganglia learned, through trial and error, to run away from ducks and run toward rabbits. In this case, the basal ganglia will have opposite reactions to seeing the duck/rabbit depending on what pattern gets sent to it from the neocortex. If the aPFC had previously imagined seeing a rabbit and running toward it, then it can control the basal ganglia’s choices by using attention to ensure that when the rat sees this ambiguous picture, it sees a rabbit, not a duck.
控制正在进行的行为通常还需要工作记忆——在没有任何感官线索的情况下维持表象。许多想象的路径和任务都涉及等待。例如，当啮齿动物在树丛中寻找坚果时，它必须记住它已经采摘了哪些树。这项任务被证明需要 aPFC。如果你在这些延迟期间抑制啮齿动物的 aPFC，啮齿动物就会失去根据记忆执行此类任务的能力。并且在执行此类任务期间，aPFC 表现出“延迟活动”，即使在没有任何外部线索的情况下也保持激活状态。这些任务需要 aPFC，因为工作记忆的功能与注意力和计划相同——它是调用内部模拟。工作记忆——将某些东西记在脑子里——只是你的 aPFC 试图不断重新调用内部模拟，直到你不再需要它。
Controlling ongoing behavior often also requires working memory—the maintenance of representations in the absence of any sensory cues. Many imagined paths and tasks involve waiting. For example, when a rodent forages among trees for nuts, it must remember which trees it has already foraged. This is a task shown to require the aPFC. If you inhibit a rodent’s aPFC during these delay periods, rodents lose their ability to perform such tasks from memory. And during such tasks, the aPFC exhibits “delay activity,” remaining activated even in the absence of any external cues. These tasks require the aPFC because working memory functions in the same way as attention and planning—it is the invoking of an inner simulation. Working memory—holding something in your head—is just your aPFC trying to keep re-invoking an inner simulation until you no longer need it.
除了计划、注意力和工作记忆之外，aPFC 还可以更直接地控制正在进行的行为：它可以抑制杏仁核。aPFC 会向杏仁核周围的抑制神经元投射。在实现想象中的计划时，aPFC 可以尝试阻止杏仁核触发自己的接近和回避反应。这是心理学家所说的行为抑制、意志力和自我控制的进化开端：我们时刻渴望的东西（由杏仁核和基底神经节控制）和我们知道的更好选择（由 aPFC 控制）之间的持续紧张关系。在意志力强大的时候，你可以抑制杏仁核驱动的渴望。在软弱的时候，杏仁核就会获胜。这就是为什么人们在疲倦或有压力时会变得更加冲动的原因——aPFC 的运行需要消耗大量的能量，因此如果你感到疲倦或有压力，aPFC 抑制杏仁核的效果就会大大降低。
In addition to planning, attention, and working memory, the aPFC can also control ongoing behavior more directly: It can inhibit the amygdala. There is a projection from the aPFC to inhibitory neurons surrounding the amygdala. During the fulfillment of an imagined plan, the aPFC can attempt to prevent the amygdala from triggering its own approach and avoidance responses. This was the evolutionary beginning of what psychologists call behavioral inhibition, willpower, and self-control: the persistent tension between our moment-to-moment cravings (as controlled by the amygdala and basal ganglia) and what we know to be a better choice (as controlled by the aPFC). In moments of willpower, you can inhibit your amygdala-driven cravings. In moments of weakness, the amygdala wins. This is why people become more impulsive when tired or stressed—the aPFC is energetically expensive to run, so if you are tired or stressed, the aPFC will be much less effective at inhibiting the amygdala.
总结一下：计划、注意力和工作记忆都由 aPFC 控制，因为从原则上讲，这三个都是同一件事。它们都是大脑试图选择要呈现的模拟的不同表现。aPFC 如何“控制”行为？这里提出的想法是，它本身并不控制行为；它试图说服基底神经节通过间接地向大脑展示一个更好的选择，并过滤哪些信息进入基底神经节，大脑皮层就会做出正确的选择。aPFC 不是通过告诉而是通过展示来控制行为。
To summarize: Planning, attention, and working memory are all controlled by the aPFC because all three are, in principle, the same thing. They are all different manifestations of brains trying to select what simulation to render. How does the aPFC “control” behavior? The idea presented here is that it doesn’t control behavior per se; it tries to convince the basal ganglia of the right choice by vicariously showing it that one choice is better and by filtering what information makes it to the basal ganglia. The aPFC controls behavior not by telling but by showing.
比较哺乳动物与蜥蜴等其他脊椎动物在需要抑制反射反应以做出“更聪明”选择的任务中的表现，可以看出这样做的好处。如果你把一只蜥蜴放在迷宫里，试着训练它走向红灯去寻找诱人的食物，避开提供不吸引人的食物的绿灯，这需要蜥蜴需要数百次尝试才能学会这个简单的任务。蜥蜴对绿光的固有偏好需要很长时间才能被训练出来。由于没有大脑皮层来暂停并间接考虑选择，蜥蜴学习这项任务的唯一方法就是通过无休止的真实尝试和错误。相比之下，老鼠学会抑制它们的固有反应要快得多，这一优势已经消失如果你损坏了老鼠的 aPFC。
The benefit of this can be seen when comparing the performance of mammals to other vertebrates like lizards on tasks that require inhibiting reflexive responses in favor of “smarter” choices. If you put a lizard in a maze and try to train it to go toward a red light to get appealing food and avoid a green light which offers unappealing food, it takes lizards hundreds of trials to learn this simple task. The hardwired preference of lizards toward green light takes a long time to be untrained. Without a neocortex to pause and vicariously consider options, the only way lizards learn this task is through endless real trial and error. In contrast, rats learn to inhibit their hardwired responses much more rapidly, an advantage that disappears if you damage a rat’s aPFC.
早期哺乳动物有能力间接探索他们内心的世界模型，根据想象的结果做出选择，并坚持一旦选定的想象计划。他们可以灵活地决定何时模拟未来，何时使用习惯；他们聪明地选择要模拟的内容，克服了搜索问题。他们是我们的第一批有目标的祖先。
Early mammals had the ability to vicariously explore their inner model of the world, make choices based on imagined outcomes, and stick to the imagined plan once chosen. They could flexibly determine when to simulate futures and when to use habits; and they intelligently selected what to simulate, overcoming the search problem. They were our first ancestors to have goals.
14
14
洗碗机器人的秘密
The Secret to Dishwashing Robots
想象一下以下场景。当你拿着这本书时，你的右手开始抽筋。你毫不费力地调整好每个手指的位置，使之完美地平衡手中的书本，但随着你失去对右臂肌肉的控制，这些手指开始萎缩。你意识到自己再也无法单独控制每根手指；你只能同时移动所有手指来张开或握紧手，你的手从灵巧的工具变成了不协调的爪子。几分钟之内，你甚至无法用右手抓住这本书，你的手臂变得虚弱无力，无法抬起。这就是中风（大脑某个区域血流减少）发生在运动皮层时的感觉。这种情况会使患者失去精细运动技能，甚至可能导致瘫痪。
IMAGINE THE FOLLOWING. As you are holding this book your right hand begins to cramp up. The specific placement of each individual finger that you have effortlessly configured to perfectly balance the book in your hand begins to wilt as you lose control of the muscles in your right arm. You realize you can no longer control each finger individually; you can only open or close your hand with all your fingers moving at once, your hand transforming from a dexterous tool to an uncoordinated claw. Within minutes you can no longer even grasp the book with your right hand at all, and your arm becomes too weak to lift. This is what the experience of a stroke—the loss of blood flow to a region of the brain—feels like when it occurs in the motor cortex. Such a condition deprives its patients of fine motor skills and can even cause paralysis.
运动皮层是大脑皮层中位于额叶皮层边缘的薄带。运动皮层构成了整个身体的地图，每个区域控制特定肌肉的运动。虽然整个运动皮层涵盖了身体的每个部位，但它并没有为每个身体部位分配相同的空间。相反，它将大量空间分配给动物能够熟练控制运动的身体部位（灵长类动物的嘴和手），而分配给它们无法很好控制的区域（如脚）的空间则少得多。运动皮层中的这张地图在相邻的体感皮层中得到反映——体感皮层是大脑皮层中处理体感信息（如皮肤中的触觉传感器和来自肌肉的本体感受信号）的区域。
The motor cortex is a thin band of neocortex on the edge of the frontal cortex. Motor cortex makes up a map of the entire body, with each area controlling movements of specific muscles. While the entire motor cortex accounts for every part of the body, it does not dedicate equal space to each body part. Instead, it dedicates lots of space to the parts of the body that animals have skilled motor control over (in primates, this is the mouth and hands) and much less space to areas that they can’t control well (like the feet). This map in the motor cortex is mirrored in the adjacent somatosensory cortex—the region of the neocortex that processes somatosensory information (such as touch sensors in the skin and proprioceptive signals from muscles).
图 14.1：人类的运动皮层
Figure 14.1: The motor cortex of humans
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
在人类中，运动皮层是控制运动的主要系统。刺激运动皮层的特定区域不仅可以引起相应身体部位的运动，而且损坏运动皮层的相同区域会导致相同身体部位瘫痪。中风患者的运动障碍几乎总是源于运动皮层区域的损伤。在黑猩猩、猕猴和狐猴中，运动皮质损伤也有这种影响。在灵长类动物中，来自运动皮质的神经元会直接向脊髓发送投射来控制运动。所有这些都可以得出这样的结论：运动皮质是运动指令的来源；它是运动的控制者。
In humans, the motor cortex is the primary system for controlling movement. Not only does stimulating specific areas of the motor cortex elicit movement from the corresponding body part, but damaging those same areas of the motor cortex creates paralysis in that same body part. The movement deficits of stroke patients almost always derive from damage to areas of the motor cortex. In chimpanzees, macaques, and lemurs, motor cortex damage also has this effect. In primates, neurons from the motor cortex send a projection directly to the spinal cord to control movement. All this leads to the conclusion that the motor cortex is the locus of motor commands; it is the controller of movement.
但这个想法有三个问题。首先，运动皮层中的新皮层柱与大脑皮层的其他区域。如果我们相信大脑皮层实现了一个生成模型，该模型试图解释其输入，并使用这些解释进行预测，那么我们必须解释如何将其重新用于产生运动命令。
But there are three problems with this idea. First, the neocortical columns in the motor cortex have the same microcircuitry as other areas of the neocortex. If we believe the neocortex implements a generative model that tries to explain its inputs and uses these explanations to make predictions, then we must account for how this could have been repurposed to produce motor commands.
其次，有些哺乳动物没有运动皮层，但它们显然可以正常移动。正如上一章所讨论的，大多数进化神经科学家认为，第一批哺乳动物的额叶皮层中唯一存在的部分是无颗粒前额叶皮层 (aPFC)；没有运动皮层。运动皮层出现在数十年前第一批哺乳动物出现后几百万年，只有胎盘谱系的哺乳动物才进化成为今天的啮齿动物、灵长类动物、狗、马，蝙蝠、大象和猫。
Second, some mammals don’t have a motor cortex, and they can clearly move around normally. As discussed in the previous chapter, most evolutionary neuroscientists believe that the only part of the frontal cortex that was present in the first mammals was the agranular prefrontal cortex (aPFC); there was no motor cortex. The motor cortex emerged tens of millions of years after the first mammals and only in the placental lineage—the mammals that would go on to become today’s rodents, primates, dogs, horses, bats, elephants, and cats.
Figure 14.2: Leading theory on evolution of the motor cortex
原创艺术由 Mesa Schumacher 创作
Original art by Mesa Schumacher
“运动皮层等于运动指令”这一观点的第三个问题是，运动皮层损伤导致的瘫痪是灵长类动物独有的；大多数运动皮层受损的哺乳动物不会患有这种瘫痪的老鼠和猫运动皮质受损后，仍然可以正常行走、狩猎、进食和活动。运动皮质显然不是早期哺乳动物的运动指令中心，直到后来——在灵长类动物中——它才成为运动的必需品。那么，运动皮质为何会进化呢？它最初的功能是什么？灵长类动物发生了哪些变化？
The third problem with the “motor cortex equals motor commands” idea is that the paralysis caused by motor cortex damage is unique to primates; most mammals with motor cortex damage do not suffer from such paralysis. Rats and cats with damaged motor cortices can still walk, hunt, eat, and move around just fine. The motor cortex was clearly not the locus of motor commands in early mammals, and it was only later—in primates—that it became required for movement. So why did the motor cortex evolve? What was its original function? What changed with primates?
预测，而不是命令
Predictions, Not Commands
主动推理理论的先驱卡尔·弗里斯顿 (Karl Friston) 对运动皮层提出了另一种解释。虽然主流观点一直认为运动皮层会发出运动指令，告诉肌肉该做什么，但弗里斯顿却颠覆了这一观点：也许运动皮层不会发出运动指令，而是发出运动预测。也许运动皮层一直处于观察状态身体运动发生在附近的躯体感觉皮层（这就是为什么运动皮层和躯体感觉皮层之间有如此精致的镜像），然后试图解释这种行为，并利用这些解释来预测动物下一步会做什么。也许只是调整了线路，让运动皮层的预测流向脊髓并控制我们的运动——换句话说，运动皮层的线路是为了让它的预测成真。
Karl Friston, the pioneer of the theory of active inference, offers an alternative interpretation of the motor cortex. While the prevailing view has always been that the motor cortex generates motor commands, telling muscles exactly what to do, Friston flips this idea on its head: Perhaps the motor cortex doesn’t generate motor commands but rather motor predictions. Perhaps the motor cortex is in a constant state of observing the body movements that occur in the nearby somatosensory cortex (hence why there is such an elegant mirror of motor cortex and somatosensory cortex) and then tries to explain the behavior and use these explanations to predict what an animal will do next. And perhaps the wiring is merely tweaked so that motor cortex predictions flow to the spinal cord and control our movement—in other words, the motor cortex is wired to make its predictions come true.
根据这一理论，运动皮质的运作方式与无颗粒前额皮质的运作方式相同。不同之处在于，aPFC 学会预测导航路径的运动，而运动皮质学会预测特定身体部位的运动。aPFC 会预测动物会向左转；运动皮质会预测动物会将左爪准确地放在平台上。
By this account, the motor cortex operates the same way the agranular prefrontal cortex operates. The difference is that the aPFC learns to predict movements of navigational paths, whereas the motor cortex learns to predict movements of specific body parts. The aPFC will predict that an animal will turn left; the motor cortex will predict that the animal will place its left paw exactly on a platform.
这就是“具身化”的一般概念——大脑皮层的各个部分，如运动皮层和躯体感觉皮层，都拥有动物身体的完整模型，可以随着时间的推移进行模拟、操控和调整。弗里斯顿的想法解释了如何重新利用大脑皮层微电路来产生特定的身体动作。
This is the general idea of “embodiment”—parts of the neocortex, such as the motor cortex and somatosensory cortex, have an entire model of an animal’s body that can be simulated, manipulated, and adjusted as time unfolds. Friston’s idea explains how the neocortical microcircuit could be repurposed to produce specific body movements.
但如果大多数哺乳动物在没有运动皮层的情况下也能正常活动，那么它最初的功能是什么？如果 aPFC 能够规划导航路线，那么运动皮层又能做什么呢？
But if most mammals can move around normally with no motor cortex, then what was its original function? If the aPFC enables the planning of navigational routes, what did the motor cortex enable?
非灵长类哺乳动物（如啮齿类和猫）的运动皮质受损会产生两种影响。首先，动物无法完成熟练的动作，例如小心翼翼地将爪子放在细树枝上、穿过小孔抓取食物、跨过视线之外的障碍物，或将脚放在不平整的小平台。其次，非灵长类哺乳动物在学习从未做过的新动作序列时会受到损害。例如，一只训练有素的老鼠，在完成特定动作序列后，只有在其运动皮层受损的情况下，才能完成这一序列。如果在老鼠接受这项任务的训练之前，损坏了它的运动皮层，它就无法学习杠杆顺序。
Damage to the motor cortex in nonprimate mammals such as rodents and cats has two effects. First, animals become impaired at performing skilled movements, such as carefully placing a paw on a thin branch, reaching through a small hole to grasp a morsel of food, stepping over an obstacle once it is out of sight, or placing a foot on a small unevenly placed platform. Second, nonprimate mammals become impaired at learning new sequences of movements that they have never performed before. For example, a rat trained to perform a specifically orchestrated sequence of movements can perform this sequence only if its motor cortex was damaged after the task was already well learned. If you damage a rat’s motor cortex before it is trained on this task, it becomes unable to learn the lever sequence.
这表明运动皮层最初不是运动指令的场所，而是运动规划的场所。当动物必须执行仔细的动作时动作——将爪子放在小平台上或跨过视线之外的障碍物——它必须提前在心理上规划和模拟其身体动作。这解释了为什么运动皮层对于学习新的复杂动作是必要的，但对于执行已经学会的动作却不是必要的。当动物学习新动作时，运动皮层模拟会间接训练基底神经节。一旦动作已经学会，就不再需要运动皮层了。
This suggests that the motor cortex was originally not the locus of motor commands but of motor planning. When an animal must perform careful movements—placing a paw on a small platform or stepping over an out-of-sight obstacle—it must mentally plan and simulate its body movements ahead of time. This explains why the motor cortex is necessary for learning new complex movements but not for executing well-learned ones. When an animal is learning a new movement, the motor cortex simulations vicariously train the basal ganglia. Once a movement is well learned, the motor cortex is no longer needed.
Figure 14.3: Cats struggle to perform planned movements after motor cortex damage
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
记录运动皮层中的活动支持了这一想法。在非灵长类哺乳动物中，运动皮层最活跃的不是一般的运动，而是特定的需要规划的动作。与动物模拟运动的想法一致，运动皮层和躯体感觉皮层会在即将到来的精确运动之前被激活，即使没有看到障碍物，而只是已知存在。并且这种活动会持续到动物完成它大概是计划中的运动。
Recording activity in the motor cortex supports this idea. In nonprimate mammals, the motor cortex is most activated not by movement in general but specifically by movements that require planning. Consistent with the idea that animals are simulating movements, the motor cortex and somatosensory cortex get activated in advance of an upcoming precision movement, even if obstacles are not seen but simply known to be there. And this activity maintains itself until the animal has completed the movement it presumably planned.
有大量证据表明，在人类中，运动前皮质和运动皮质都是通过做动作和想象动作来激活的：例如，让一个人想象自己走路，他的腿部区域运动皮层被激活。这种想象运动和实际运动的神经基础结构交织在一起，不仅可以在大脑记录中观察到，也可以在物理实验室实验中观察到。让一个人坐在椅子上，让他们只做简单的保持直立姿势。然后播放任意句子的录音。当她听到“我起床，穿上拖鞋，去洗手间”这样的句子时，她的姿势会变得更糟，但当她听到句子时则不会与运动无关。仅仅听到这些句子就会激活改变姿势的内在模拟，从而影响她的实际姿势。当然，这种内在模拟也有好处（它不会只是打乱我们的姿势）：在脑海中演练运动技能可以大大提高说话、高尔夫挥杆、甚至外科手术操作。
In humans there is plenty of evidence that the premotor and motor cortices are activated both by doing movements and by imagining movements: For example, have someone think about walking, and the leg area of the motor cortex becomes activated. This intertwining of the neurological infrastructure for imagined movements and actual movements can be observed not only in brain recordings, but also in physical lab experiments. Sit a human down in a chair and ask them to do nothing more than simply maintain their upright posture. Then play audio recordings of arbitrary sentences. Her posture gets worse when she hears sentences like “I get up, put on my slippers, go to the bathroom,” but not when she hears sentences unrelated to movement. Just hearing those sentences activated an inner simulation of a changing posture, which affected her actual posture. Of course, this inner simulation also comes with benefits (it doesn’t just mess up our posture): Mental rehearsal of motor skills substantially increases performance across speaking, golf swings, and even surgical maneuvers.
运动皮层的感觉运动规划能力使早期哺乳动物能够学习和执行精确的动作。当将哺乳动物的运动技能与爬行动物的运动技能进行比较时，很明显哺乳动物在精细运动技能方面具有独特的能力。老鼠可以捡起种子并熟练地将其打开。老鼠、松鼠和猫是极其熟练的爬树者，它们毫不费力地将四肢放在精确的位置以确保它们不会掉落。松鼠和猫可以计划和执行极其准确和精确的跨平台跳跃。如果你曾经养过蜥蜴或乌龟作为宠物，你就会知道这些技能不是大多数爬行动物的专长。事实上，研究蜥蜴越过障碍物的研究揭示了令人惊讶的马虎整个事件都是。它们不会预测障碍物，也不会改变前肢的位置绕过平台。鉴于它们似乎无法提前规划运动，也许并不奇怪，很少有爬行动物生活在树上，而且那些生活在树上的爬行动物移动缓慢，这与树栖哺乳动物快速熟练的奔跑和跳跃形成了鲜明对比。
The motor cortex’s skill in sensorimotor planning enabled early mammals to learn and execute precise movements. When comparing the motor skills of mammals to that of reptiles, it is quite clear that mammals are uniquely capable when it comes to fine motor skills. Mice can pick seeds up and skillfully break them open. Mice, squirrels, and cats are incredibly skillful tree climbers, effortlessly placing their limbs in precise places to ensure they don’t fall. Squirrels and cats can plan and execute extremely accurate and precise jumps across platforms. If you have ever had a pet lizard or a turtle, you know that such skills are not the province of most reptiles. In fact, studies that examined lizards running over obstacles revealed how surprisingly sloppy the whole affair is. They don’t anticipate obstacles or modify their forelimb placements to get around platforms. Given their seeming inability to plan movements in advance, it is perhaps unsurprising that few reptiles live in trees and that those that do move slowly, in contrast to the rapid skillful running and jumping of arboreal mammals.
目标层次：模拟与自动化的平衡
A Hierarchy of Goals: A Balance of Simulation and Automation
所有这些是如何协同工作的？早期胎盘哺乳动物的额叶新皮质被组织成一个层次结构。层次结构的顶部是无颗粒前额皮质，高级目标是根据杏仁核和下丘脑的激活构建的。aPFC 可能会产生“喝水”或“吃食物”之类的意图。然后，aPFC 将这些目标传播到附近的额叶区域（运动前皮质），该区域构建子目标，并进一步传播这些子目标，直到它们到达运动皮质，然后运动皮质构建子子目标。在运动皮层是这些子子目标，它们可以简单到“把我的食指放在这里，把我的拇指放在这里”。
How does all this work together? The frontal neocortex of early placental mammals was organized into a hierarchy. At the top of the hierarchy was the agranular prefrontal cortex, where high-level goals are constructed based on amygdala and hypothalamus activation. The aPFC might generate an intent like “drink water” or “eat food.” The aPFC then propagates these goals to a nearby frontal region (the premotor cortex), which constructs subgoals and propagates these subgoals further until they reach the motor cortex, which then constructs sub-subgoals. The intent modeled in the motor cortex are these sub-subgoals, which can be as simple as “Position my index finger here and my thumb over here.”
图 14.4：早期胎盘哺乳动物的运动层次
Figure 14.4: The motor hierarchy in early placental mammals
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
这种层次结构通过将精力分散到许多不同的新皮质柱上，实现了更高效的处理。aPFC 不必担心实现其目标所需的具体动作；它只需担心高级导航路径。同样，这使得运动皮质不必担心行为的高级目标，而只需担心实现特定的低级运动目标（拿起杯子或演奏特定和弦）。
This hierarchy enables more efficient processing by distributing effort across many different neocortical columns. The aPFC doesn’t have to worry about the specific movements necessary to achieve its goals; it must worry only about high-level navigational paths. Similarly, this allows the motor cortex not to have to worry about the high-level goal of the behavior and worry only about accomplishing specific low-level movement goals (picking up a cup or playing a specific chord).
基底神经节与额叶皮质形成连接环路，aPFC 连接到基底神经节的前部区域（然后通过丘脑连接回 aPFC），运动皮质连接到基底神经节的后部区域（然后通过丘脑的另一个区域连接回运动皮质）。这些环路连接得如此优雅和特别，以至于很难抗拒尝试逆向工程来了解它们的作用。
The basal ganglia makes loops of connectivity with the frontal cortex, with the aPFC connecting to the front region of the basal ganglia (which then connects back to aPFC through the thalamus), and the motor cortex connecting to the back region of the basal ganglia (which then connects back to the motor cortex through a different region of the thalamus). These loops are so elegantly and particularly wired that it is hard to resist trying to reverse-engineer what they are doing.
神经科学家的主流观点是，这些是旨在管理不同层次的运动层次的子系统。基底神经节的前部会自动将刺激与高级目标联系起来。正是它产生了渴望：你回到家，闻到通心粉的香味，突然间，你就有了吃一些通心粉的任务。当吸毒者看到引起吸毒渴望的刺激时，基底神经节的前部就会极度活跃。然而，aPFC 会让你停下来思考如果你真的想追求这些渴望（“那我们的饮食怎么样？”）。基底神经节的后部会自动将刺激与低级目标联系起来，比如特定的身体动作。这就是产生自动熟练动作的原因。另一方面，运动皮层会让你停下来，提前计划好你的确切动作。
The leading view among neuroscientists is that these are subsystems designed to manage different levels of the motor hierarchy. The front part of the basal ganglia automatically associates stimuli with high-level goals. It is what generates cravings: You come home and smell rigatoni, and suddenly you are on a mission to eat some. Drug addicts show extreme activations of this front part of the basal ganglia when they see stimuli that create drug cravings. The aPFC, however, is what makes you pause and consider if you actually want to pursue these cravings (“What about our diet?”). The back part of the basal ganglia automatically associates stimuli with low-level goals, such as specific body movements. It is what generates automatic skilled movements. The motor cortex, on the other hand, is what makes you pause and plan out your exact movements ahead of time.
|
高层目标 HIGH-LEVEL GOALS |
低级目标 LOW-LEVEL GOALS |
模拟 SIMULATION |
无颗粒前额皮质 Agranular Prefrontal Cortex 模拟导航路径 Simulates navigational paths 问“我想要通心粉还是想节食？” Asks “do I want rigatoni or would I rather diet?” 损坏导致航行路线规划受阻 Damage causes impairments in planning navigational routes |
运动皮层 Motor Cortex 模拟身体动作 Simulates body movements 问“我该如何配置我的手指来弹奏我刚刚在吉他上学到的 C 和弦？” Asks “how do I configure my fingers to play this C chord that I just learned on guitar?” 损伤会导致学习新运动技能和执行精细运动技能受损 Damage causes impairments in learning new motor skills and executing fine motor skills |
自动化 AUTOMATION |
基底神经节前部 Front part of basal ganglia 受到刺激后自动追求高水平目标 Automatic pursuit of a high-level goal in response to stimulus 产生习惯性渴望 Produces habitual cravings |
基底神经节后部 Back part of basal ganglia 响应刺激而自动执行运动技能 Automatic execution of motor skill in response to stimulus 产生习惯性运动反应 Produces habitual motor responses 损伤会导致学习技能的执行受损，并损害运动习惯的形成 Damage causes impairments in executing learned skills, and impairs motor habit formation |
任何级别的目标，无论是高级目标还是低级目标，在额叶新皮质中都有一个自我模型，在基底神经节中则有一个无模型系统。新皮质为训练提供了一个较慢但更灵活的系统，而基底神经节为训练好的路径和动作提供了一个较快但不太灵活的版本。
Any level of goal, whether high-level or low-level goals, has both a self model in the frontal neocortex and a model-free system in the basal ganglia. The neocortex offers a slower but more flexible system for training, and the basal ganglia offers a faster but less flexible version for well-trained paths and movements.
有大量证据表明这种运动层次结构。记录显示，aPFC 中的神经元对高层次目标敏感，而运动前区和运动皮层中的神经元对逐渐增加的目标敏感。较低级别的子目标。学习一种新行为最初会激活运动层次的所有级别，但随着行为成为自动行为，它只会激活较低级别的子目标。层次结构中的级别。如果你损坏了破坏大鼠运动层次结构（aPFC 或基底神经节前部）会使它们对高级目标不那么敏感（它们会继续推动杠杆，即使它们不再需要杠杆生产的食物）。相反，如果破坏运动层次结构的低级部分，大鼠就会对高级目标变得更加敏感，并且它们很难养成运动习惯（例如，无论老鼠按了多少次杠杆，它们都不会养成推动杠杆的习惯他们所经历的考验）。
There is plenty of evidence for such a motor hierarchy. Recordings have shown that neurons in the aPFC are sensitive to high-level goals, whereas those in the premotor and motor cortex are sensitive to progressively lower-level subgoals. Learning a new behavior activates all levels of the motor hierarchy at first, but as the behavior becomes automatic, it activates only lower levels in the hierarchy. If you damage high-level parts of the motor hierarchy (aPFC or the front part of the basal ganglia) in rats, it makes them less sensitive to high-level goals (they will keep pushing a lever even though they no longer want the food it produces). In contrast, if you damage low-level parts of the motor hierarchy, rats become more sensitive to high-level goals, and they struggle to create motor habits (e.g., rats won’t develop a habit for pushing a lever no matter how many trials they go through).
虽然 aPFC 受损会使动物失去意图，就像我们在患者 L 身上看到的那样，但运动前皮层部分受损似乎会切断这些意图从高级目标到具体动作的正确流动。这可能会导致“异肢综合症”：患者会声称他们身体的某些部位在移动不受控制地自行移动。这种外来移动的迹象也出现在啮齿类动物身上，运动前皮质受损。这种损伤还会导致所谓的“利用行为”或“场依赖行为”，患者会执行没有任何明确目标的运动序列：他们会从空杯子里喝水，即使不去任何地方也会穿上别人的夹克，用铅笔乱涂乱画，或者做任何其他行为附近的刺激暗示了这一点。所有这些都是层次结构被打破的结果——运动皮层的部分区域现在不受来自 aPFC 自上而下流经运动前皮层的意图的约束，因此运动皮层独立地为运动序列设定了低级目标。
While damage to the aPFC deprives animals of their intentions, as we saw with patient L, damage to parts of the premotor cortex seems to disconnect the proper flow of these intentions from high-level goals to specific movements. This can cause “alien limb syndrome”: patients will claim that certain parts of their bodies are moving on their own without their control. Signs of such alien movements are also seen in rodents with damage to the premotor cortex. Such damage also causes what is called “utilization behavior” or “field-dependent behavior,” where patients will execute motor sequences devoid of any clear goal: They will drink from empty cups, put on other people’s jackets even though they aren’t going anywhere, scribble with pencils, or do any other behavior that nearby stimuli suggest. All this is the result of a broken hierarchy—parts of the motor cortex are now unconstrained by top-down intentions from the aPFC flowing through the premotor cortex, and hence the motor cortex independently sets low-level goals for motor sequences.
还有大量证据支持这一观点：额叶新皮质是模拟的中心，而基底神经节是自动化的中心。损坏动物的运动皮质会损害运动规划和学习新动作的能力，但不会影响执行训练有素的动作（因为基底神经节的后部已经学会了这些动作）。同样，损坏动物的 aPFC 会损害路径规划和学习新路径的能力，但不会影响执行训练有素的路径的能力。
There is also plenty of evidence for the idea that the frontal neocortex is the locus of simulation, while the basal ganglia is the locus of automation. Damaging an animal’s motor cortex impairs movement planning and learning new movements but not the execution of well-trained movements (because the back part of the basal ganglia already learned them). Similarly, damaging an animal’s aPFC impairs path planning and learning new paths but not the execution of well-trained paths.
此外，基底神经节的前部表现出自动选择要追求的线索的所有迹象（即自动化高级行为）。当你看到一个引起渴望的线索时，大脑中最活跃的部分就是基底神经节的前部。那些试图抑制渴望的人会表现出额叶区域（如 aPFC）的额外激活（模拟负面后果并试图训练基底神经节做出更艰难的选择）。事实上，损伤基底神经节前部是一种有效的（尽管极具争议且道德性值得怀疑的）药物成瘾治疗方法。海洛因成瘾者的复发率高得离谱；有人估计高达 90%。中国的一项研究选取了最严重的海洛因成瘾者，损伤了他们的基底神经节前部。复发率率下降到42%。人们失去了追寻线索和产生渴望的自动行为（当然，这样的手术也有很多副作用）。
Further, the front part of the basal ganglia demonstrates all the signs of an automatic selection of cues to pursue (i.e., automated high-level behaviors). When you see a cue that creates a craving, the part of the brain that is most activated is this front part of the basal ganglia. Those who try to inhibit their cravings show additional activation of frontal areas like the aPFC (simulating the negative consequences and trying to train the basal ganglia to make the harder choice). In fact, lesioning the front part of the basal ganglia is an effective (although highly controversial and questionably ethical) treatment for drug addiction. The relapse rate for heroin addicts is absurdly high; some estimate as high as 90 percent. One study in China took the most severe heroin addicts and lesioned the front part of their basal ganglia. The relapse rate dropped to 42 percent. People lose the automatic behavior of pursuing cues and generating cravings (of course, there are also many side effects of such a surgery).
完整且功能良好的运动层次结构将使早期胎盘哺乳动物的行为具有惊人的灵活性；动物可以在 aPFC 中设定高级目标，而运动层次结构的较低级别区域可以灵活地应对出现的任何障碍。追逐远处水源的哺乳动物可以随着事件的展开不断更新其子目标——运动前皮层可以通过选择新的运动序列来应对意外障碍，运动皮层甚至可以调整肢体最细微的特定运动，所有这些都是为了实现一个共同的目标。
An intact and well-functioning motor hierarchy would have made the behavior of early placental mammals impressively flexible; animals could set high-level goals in the aPFC while lower-level areas of the motor hierarchy could flexibly respond to whatever obstacles present themselves. A mammal pursuing faraway water could continuously update its subgoals as events unfolded—the premotor cortex could respond to surprising obstacles by selected new movement sequences, and the motor cortex could adjust even the subtlest of specific movements of limbs, all in the name of a common goal.
洗碗机器人的秘密就在于哺乳动物的运动皮层和更广泛的运动系统。正如我们尚不了解大脑皮层微电路如何精确模拟感官输入一样，我们也尚不了解运动皮层如何如此灵活和准确地模拟和规划精细的身体动作，以及它如何在运动过程中不断学习。
The secret to dishwashing robots lives somewhere in the motor cortex and the broader motor system of mammals. Just as we do not yet understand how the neocortical microcircuit renders an accurate simulation of sensory input, we also do not yet understand how the motor cortex simulates and plans fine body movements with such flexibility and accuracy and how it continuously learns as it goes.
但如果我们以过去几十年的经验为指导，机器人专家和人工智能研究人员可能会在不久的将来找到答案。事实上，机器人技术正在快速进步。二十年前，我们几乎无法让四足机器人保持直立平衡，而现在我们拥有了可以在空中翻转的人形机器人。
But if we use the past few decades as our guide, roboticists and AI researchers will likely figure this out, perhaps in the near future. Indeed, robotics are improving at a rapid pace. Twenty years ago, we could barely get a four-legged robot to balance itself upright, and now we have humanoid robots that can do flips in the air.
如果我们成功制造出具有与哺乳动物相似的运动系统的机器人，它们将具有许多理想的特性。这些机器人将自动学习新的复杂技能。它们将实时调整自己的动作以应对世界的扰动和变化。我们将为它们设定高层次的目标，它们将能够找出实现目标所需的所有子目标。当它们尝试学习某项新任务时，它们会缓慢而谨慎地模拟每个任务他们在行动之前会先做出身体动作，但随着他们的进步，行为会变得更加自动化。在他们的一生中，他们学习新技能的速度会随着他们将以前学到的低级技能重新应用到新体验的高级目标中而增加。如果他们的大脑像哺乳动物的大脑一样工作，他们就不需要大型超级计算机来完成这些任务。事实上，整个人类大脑运作所需的能量与灯泡的能量大致相同。
If we successfully build robots with motor systems similar to those of mammals, they will come along with many desirable properties. These robots will automatically learn new complex skills on their own. They will adjust their movements in real time to account for perturbations and changes in the world. We will give them high-level goals, and they will be able to figure out all the subgoals necessary to achieve it. When they try to learn some new task, they will be slow and careful as they simulate each body movement before they act, but as they get better, the behavior will become more automatic. Over the course of their lifetimes, the speed with which they learn new skills will increase as they reapply previously learned low-level skills to newly experienced higher-level goals. And if their brains work at all like mammal brains, they will not require massive supercomputers to accomplish these tasks. Indeed, the entire human brain operates on about the same amount of energy as a lightbulb.
也可能不是。也许机器人专家会以一种非常非哺乳动物的方式让这一切发挥作用——也许机器人专家会在不逆向工程人类大脑的情况下解决这一切。但就像鸟翼是飞行可能性的存在证明——这是人类努力追求的目标——哺乳动物的运动技能是我们希望有朝一日在机器中建立的运动技能类型的存在证明，而运动皮层和周围的运动层次是大自然关于如何让这一切发挥作用的线索。
Or maybe not. Perhaps roboticists will get all this to work in a very nonmammalian way—perhaps roboticists will figure it all out without reverse-engineering human brains. But just as bird wings were an existence proof for the possibility of flight—a goal for humans to strive for—the motor skills of mammals are our existence proof for the type of motor skills we hope to build into machines one day, and the motor cortex and the surrounding motor hierarchy are nature’s clues about how to make it all work.
突破 3 号总结：模拟
Summary of Breakthrough #3: Simulating
早期哺乳动物大脑中出现的主要新结构是大脑皮层。大脑皮层带来了模拟能力——这是我们进化史上的第三次突破。总结一下这一过程及其用途：
The primary new brain structure that emerged in early mammals was the neocortex. With the neocortex came the gift of simulation—the third breakthrough in our evolutionary story. To summarize how this occurred and how it was used:
- 感觉新皮层进化而来，创造了外部世界的模拟（世界模型）。
- Sensory neocortex evolved, which created a simulation of the external world (a world model).
- 无颗粒前额皮质（aPFC）进化而来，这是额叶新皮质的第一个区域。aPFC 模拟了动物自身的动作和内部状态（自我模型），并构建了“意图”来解释自己的行为。
- The agranular prefrontal cortex (aPFC) evolved, which was the first region of the frontal neocortex. The aPFC created a simulation of an animal’s own movements and internal states (a self model) and constructed “intent” to explain one’s own behavior.
- aPFC 和感觉大脑皮层共同作用，使早期哺乳动物能够暂停并模拟当前尚未体验到的世界的各个方面——换句话说，基于模型的强化学习。
- The aPFC and sensory neocortex worked together to enable early mammals to pause and simulate aspects of the world that were not currently being experienced—in other words, model-based reinforcement learning.
- aPFC 通过智能地选择要模拟的路径并确定何时模拟它们，以某种方式解决了搜索问题。
- The aPFC somehow solved the search problem by intelligently selecting paths to simulate and determining when to simulate them.
- 这些模拟使早期哺乳动物能够进行间接试错——模拟未来的行动并根据想象的结果决定采取哪条路径。
- These simulations enabled early mammals to engage in vicarious trial and error—to simulate future actions and decide which path to take based on the imagined outcomes.
- 这些模拟使早期哺乳动物能够进行反事实学习，从而为信用分配问题提供更先进的解决方案——使哺乳动物能够根据因果关系分配信用。
- These simulations enabled early mammals to engage in counterfactual learning, thereby offering a more advanced solution to the credit assignment problem—enabling mammals to assign credit based on causal relationships.
- 这些模拟使早期哺乳动物能够进行情景记忆，从而使哺乳动物能够回忆过去的事件和行为，并利用这些回忆来调整自己的行为。
- These simulations enabled early mammals to engage in episodic memory, which allowed mammals to recall past events and actions, and use these recollections to adjust their behavior.
- 在后来的哺乳动物中，运动皮层进化了，使哺乳动物能够计划和模拟特定的身体动作。
- In later mammals, the motor cortex evolved, enabling mammals to plan and simulate specific body movements.
我们一亿年前的哺乳动物祖先将想象作为生存的武器。他们进行了间接的尝试和错误，反事实学习和情景记忆可以超越恐龙。我们的祖先哺乳动物，比如现代猫，可以看着一组树枝并计划将爪子放在哪里。总的来说，这些古老的哺乳动物比它们的脊椎动物祖先行为更灵活，学习速度更快，并且表现出更聪明的运动技能。
Our mammalian ancestors from a hundred million years ago weaponized the imaginarium to survive. They engaged in vicarious trial and error, counterfactual learning, and episodic memory to outplan dinosaurs. Our ancestral mammal, like a modern cat, could look at a set of branches and plan where it wanted to place its paws. Together, these ancient mammals behaved more flexibly, learned faster, and performed more clever motor skills than their vertebrate ancestors.
当时的大多数脊椎动物，如现代蜥蜴和鱼类，仍然可以快速移动、记住模式、追踪时间的流逝，并通过无模型强化学习进行智能学习，但它们的动作是没有计划的。
Most vertebrates at the time, as with modern lizards and fish, could still move quickly, remember patterns, track the passage of time, and intelligently learn through model-free reinforcement learning, but their movements were not planned.
因此，思维本身并非诞生于普罗米修斯神圣工坊的粘土生物中，而是诞生于侏罗纪地球的地下小隧道和盘根错节的树木中，诞生于恐龙一亿年的掠食和我们祖先为避免灭绝而做出的绝望尝试的熔炉中。这就是我们的大脑皮层和我们对世界的内在模拟如何形成的真实故事。我们很快就会看到，下一个突破最终将从这种来之不易的超能力中出现。
And so, thinking itself was born not within the clay creatures of Prometheus’s divine workshop, but instead in the small underground tunnels and knotted trees of a Jurassic Earth, birthed from the crucible of a hundred million years of dinosaur predation and our ancestor’s desperate attempt to avoid fading into extinction. That is the real story of how our neocortex and our inner simulation of the world came into being. And as we will soon see, it was from this hard-won superpower that the next breakthrough would eventually emerge.
从某种程度上来说，这一突破是现代人工智能系统中最难逆向的突破；事实上，这一突破是我们通常不会与“智能”联系在一起的壮举，但事实上，它是我们大脑最令人印象深刻的壮举之一。
This next breakthrough has been, in some ways, the hardest breakthrough to reverse engineer in modern AI systems; indeed, this next breakthrough is a feat we don’t typically associate with “intelligence,” but is, in fact, one of our brain’s most impressive feats.
突破＃4
Breakthrough #4
心智化和第一批灵长类动物
Mentalizing and the First Primates
1500 万年前的大脑
Your brain 15 million years ago
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
15
15
政治智慧的军备竞赛
The Arms Race for Political Savvy
这件事发生在大约六千六百万年前一个平凡的日子，这一天的开始与其他日子没有任何不同。太阳从如今非洲的丛林中升起，唤醒了沉睡的恐龙，驱使我们夜行的松鼠祖先躲进了白天的藏身之处。沿着泥泞的海岸，雨水淅淅沥沥地流进浅水塘，塘里住着古老的两栖动物。潮水退去，把许多鱼和其他古老的生物都吸引到深海。天空中充满了翼龙和古老的鸟类。节肢动物和其他无脊椎动物在土壤和树木中挖洞。地球的生态找到了一种美妙的平衡，恐龙在食物链的顶端安然生存了一亿五千万年，鱼类统治海洋的时间更长，哺乳动物和其他动物也找到了各自微小但适宜生存的生态位。没有任何迹象表明这一天会与其他任何一天有任何不同，但这确实是一切都改变的一天——这是世界几乎毁灭的一天。
IT HAPPENED ON some unremarkable day around sixty-six million years ago, a day that began no different than any other. The sun rose over the jungles of today’s Africa, awakening slumbering dinosaurs and driving our nocturnal squirrel-like ancestors into their daytime hiding spots. Along the muddy seashores, rain pattered into shallow ponds filled with ancient amphibians. Tides receded, drawing the many fish and other ancient critters deep in the oceans. The skies filled with pterosaurs and ancient birds. Arthropods and other invertebrates tunneled within the soils and trees. The ecology of Earth had found a beautiful equilibrium, with dinosaurs comfortably at the top of the food chain for well over a hundred and fifty million years, fish ruling the sea for even longer, and mammals and other animals finding their respective tiny but livable niches. Nothing hinted that this day would be any different than any other, but it was indeed this day when everything changed—this was the day the world almost ended.
当然，经历过这一天的动物的具体生活故事我们已无从知晓。但我们可以推测。我们的一种松鼠类哺乳动物祖先可能正要走出洞穴，开始一夜觅食昆虫。太阳刚刚落山，天空一定变成了每隔一天晚上都会出现的紫色。但随后，黑暗从地平线上出现。一片比她见过的任何风暴都厚的乌云迅速笼罩着天空。也许她困惑地抬头看着这新奇的景象；也许她完全忽略了它。要么尽管她拥有了新皮质的智慧，她仍然无法理解正在发生的事情。
The specific life stories of any of the animals who experienced this day are, of course, lost to us. But we can speculate. One of our mammalian squirrel-like ancestors was perhaps on her way out of her burrow to begin a night of scavenging insects. As the sun was just beginning to set, the sky must have turned that purple hue it did every other evening. But then blackness emerged from the horizon. A dark cloud, thicker than any storm she had ever seen, spread rapidly over the sky. Perhaps she looked up at this novel sight with puzzlement; perhaps she ignored it completely. Either way, despite all her new neocortical smarts, she would have had no way to understand what was happening.
这不是地平线上的风暴，而是太空尘埃。就在几分钟前，在地球的另一边，一颗几英里宽的小行星撞上了地球。它抛出了巨大的地球碎片，这些碎片迅速将天空填满黑色的烟尘——这种黑暗会遮蔽太阳一百多年，造成 70% 以上的人死亡陆生脊椎动物。这就是二叠纪-三叠纪灭绝事件。
This was no storm on the horizon—this was space dust. Just a few minutes prior, on the other side of the planet, an asteroid a few miles wide had slammed into the Earth. It had sent up gargantuan chunks of earth debris that was rapidly filling the skies with dark soot—a blackness that would block out the sun for over one hundred years, killing over 70 percent of land-living vertebrates. This was the Permian-Triassic extinction.
地球历史上的许多其他灭绝事件似乎都是由自身造成的——大氧化事件是由蓝藻引起的，晚泥盆世灭绝可能是由于陆地植物过度繁殖造成的。但这一次不是生命的错，而是矛盾宇宙的偶然事件。
Many of the other extinction events in the history of Earth seem to have been self-imposed—the Great Oxygenation Event was caused by cyanobacteria, and the Late Devonian Extinction was possibly caused by overproliferation of plants on land. But this one was not a fault of life but a fluke of an ambivalent universe.
最终，经过一百多年，乌云开始消散。随着太阳重新出现，植物开始恢复失去的土地，重新填满干涸的死地。但这是一个新世界。几乎所有恐龙物种都灭绝了，只剩下一种：鸟类。虽然我们的松鼠祖先不可能知道这一点，也没有活着看到这一点，但她的后代将继承这个新的地球。随着地球的愈合，这些小型哺乳动物发现自己身处一个全新的生态游乐场。没有了恐龙捕食者，它们可以自由地探索新的生态位，进化成新的形状和大小，征服新的领地，并在食物链中找到新的立足点。
Eventually, after over a hundred years, the blackened clouds began to fade. As the sun reemerged, plants began recovering lost ground and refilling the parched dead land. But this was a new world. Almost every dinosaur species was extinct except for one: the birds. Although our squirrel-like ancestor could not have known it and had not lived to see it, her offspring would inherit this new Earth. As the Earth healed, these small mammals found themselves in a completely new ecological playground. Without their dinosaur predators, they were free to explore new ecological niches, to diversify into new shapes and sizes, to conquer new territories, and to find new footing within the food chain.
随后的时代被称为哺乳动物时代。这些早期哺乳动物的后代最终进化成为现代的马、大象、老虎和老鼠。有些甚至会重新进入海洋，成为今天的鲸鱼、海豚和海豹。有些会飞上天空，成为今天的蝙蝠。
The era that followed has been called the Era of Mammals. Descendants of these early mammals would eventually evolve into modern-day horses, elephants, tigers, and mice. Some would even reenter the sea and become today’s whales, dolphins, and seals. Some would take to the sky and become today’s bats.
我们的直系祖先在非洲的高大树木中找到了避难所。他们是第一批灵长类动物。他们从夜行动物转变为日行动物。随着体型的增大，他们进化出了对生拇指，以便抓住树枝并支撑更重的身体。为了支撑更大的体型，他们从以昆虫为主的饮食转变为以水果为主的饮食。他们群居，随着年龄的增长，他们相对不受捕食和食物竞争的困扰。最值得注意的是，他们的大脑体积膨胀到了原来的一百倍以上。
Our direct ancestors were the ones who found refuge in the tall trees of Africa. These were some of the first primates. They shifted from being night-living (nocturnal) to being day-living (diurnal). As they became larger, they developed opposable thumbs to grasp branches and hold their heavier bodies. To support their bigger size, they shifted from an insect-based diet to a fruit-based diet. They lived in groups, and as they grew, they became relatively free from predation and food competition. And most notably, their brains exploded to well over a hundred times their original size.
图 15.1：哺乳动物树
Figure 15.1: Tree of mammals
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
许多哺乳动物谱系的大脑在比例上并不比早期哺乳动物大很多，只有某些哺乳动物谱系的大脑才会显著增大，比如大象、海豚和灵长类动物。因为这本书讲述的是人类的故事，所以我们将重点介绍灵长类动物大脑变大的过程。事实上，为什么灵长类动物有这么大的大脑——特别是这么大的新皮层——这个问题自达尔文时代以来就一直困扰着科学家。早期灵长类动物的生活方式有什么需要这么大的大脑呢？
Many lineages of mammals went on to have brains not much larger (proportionally) than those of early mammals, it was only in certain lineages of mammals, such as those of elephants, dolphins, and primates, where brains dramatically expanded. Because this book is about the human story, we will focus on the journey by which the primate brain became big. Indeed, why primates have such big brains—and specifically such large neocortices—is a question that has perplexed scientists since the days of Darwin. What was it about the lifestyle of early primates that necessitated such a big brain?
社会脑假说
The Social-Brain Hypothesis
在 20 世纪 80 年代和 90 年代，许多灵长类动物学家和进化心理学家，包括尼古拉斯·汉弗莱 (Nicholas Humphrey)、弗兰斯·德瓦尔 (Frans de Waal) 和罗宾Dunbar 开始推测灵长类动物大脑的发育与一千万至三千万年前非洲丛林中猴子的生态需求无关，而是独特的社会需求的结果。他们认为这些灵长类动物拥有稳定的小型社会：个体群体长期团结在一起。科学家推测，为了维持这些独特的大型社会群体，这些个体需要独特的认知能力。他们认为，这给它们带来了更大的大脑压力。
In the 1980s and 1990s, numerous primatologists and evolutionary psychologists, including Nicholas Humphrey, Frans de Waal, and Robin Dunbar, began speculating that the growth of the primate brain had nothing to do with the ecological demands of being a monkey in the African jungles ten to thirty million years ago and was instead a consequence of the unique social demands. They argued that these primates had stable mini-societies: Groups of individuals that stuck together for long periods. Scientists hypothesized that to maintain these uniquely large social groups, these individuals needed unique cognitive abilities. This created pressure, they argued, for bigger brains.
检验这一理论的简单方法是观察世界各地的猴子和猿类的小型社会，看看它们的大脑皮层相对于大脑其他部分的大小是否与社会群体的大小相关。罗宾·邓巴 (Robin Dunbar) 就是这么做的，他的发现震惊了整个领域。这种相关性已在许多灵长类动物中得到证实：灵长类动物的大脑皮层越大，社会群体的大小就越大。其社会群体更大。
A simple test of this theory would be to look at the mini-societies of monkeys and apes around the world and see if the size of their neocortices relative to the rest of their brains correlated with their social-group size. It was Robin Dunbar who did this, and what he found shook the field. This correlation has been confirmed across many primates: the bigger the neocortex of a primate, the bigger its social group.
但猴子和猿类远非唯一的哺乳动物，更不用说动物了，群居动物。有趣的是，这种相关性并不成立对于大多数其他动物来说，情况并非如此。生活在上千头牛群中的水牛的大脑并不比孤独的驼鹿的大脑大多少。似乎不是群体大小的问题，而是早期灵长类动物所创造的特定群体类型需要更大的大脑。灵长类动物群体相对于大多数其他哺乳动物的群体有一些独特之处，我们只有通过理解群体本身的普遍驱动力才能理解这一点。
But monkeys and apes are far from the only mammals, let alone animals, that live in groups. And interestingly, this correlation does not hold for most other animals. The brain of a buffalo living in a thousand-member herd is not meaningfully bigger than the brain of a solitary moose. It isn’t group size in general but the specific type of group that early primates created that seemed to have required larger brains. There was something unique about primate groups relative to those of most other mammals, something we can understand only by understanding the general drive for grouping itself.
集体与个人之间的进化张力
The Evolutionary Tension Between the Collective and the Individual
早期哺乳动物可能比之前的羊膜动物（它们的蜥蜴类祖先）更具社会性。这些早期哺乳动物独特地生下了无助的孩子。只有当母亲建立起强大的纽带来帮助、养育和保护孩子时，这种动态才会成立。此外，哺乳动物比其他脊椎动物更多地参与游戏。即使是像老鼠这样的简单哺乳动物的后代也会互相玩耍，玩爬上和打架。这些早期的游戏行为可能起到了改善和训练年轻哺乳动物运动皮层的目的，以便在更高风险的情况下，它们不会从头开始学习。在这些早期哺乳动物中，母亲和孩子之间的这种合作期相对较短。经过一段儿童发展期后，这种纽带往往会消退，孩子和母亲会分道扬镳。许多哺乳动物就是这样，它们一生中的大部分时间都是独自度过的，比如老虎和熊。
Early mammals were likely more social than the amniotes (their lizard-like ancestors) that came before. These early mammals uniquely gave birth to helpless children. This dynamic would have been tenable only if mothers built a strong bond to help, nurture, and physically protect their children. Further, mammals engage in play much more than other vertebrates. Even the offspring of simple mammals like rats play with each other, play mounting and play fighting. These early acts of play might have served the purpose of refining and training the motor cortex of young mammals so that in higher-stake situations they wouldn’t be learning from scratch. In these early mammals, this collaborative period between mother and child was relatively short-lived. After a period of childhood development, the bond tends to fade, and the children and mothers go their separate ways. This is how it is for many mammals that spend most of their lives on their own, like tigers and bears.
但并非所有动物成年后都会像这样分开。事实上，动物最简单、最广泛使用且可能最早的集体行为就是群居，即同一物种的动物简单地聚集在一起。鱼会本能地跟随彼此的动作并在彼此附近游泳。许多食草恐龙是群居动物。当然，哺乳动物也是如此——水牛和羚羊都是群居动物。群居的主要好处是它有助于抵御掠食者。如果羚羊群中的一只羚羊瞥见附近的狮子并开始逃跑，它就会提示群中的所有其他羚羊跟随它。虽然一只孤独的羚羊很容易被捕食，但一群羚羊甚至对狮子来说都是危险的。
But not all animals separate in adulthood like this. In fact, the simplest, most widely used, and likely first collective behavior in animals was group living, whereby animals of the same species simply clustered together. Fish reflexively follow each other’s movements and swim near one another. Many herbivorous dinosaurs lived in herds. And, of course, this is seen across mammals as well—buffalo and antelope live in herds. The key benefit of group living is that it helps stave off predators. If even a single antelope in a herd catches a glimpse of a nearby lion and begins running away, it cues all the others in the herd to follow. While a lone antelope is easy prey, a herd of them can be dangerous to even a lion.
然而，群居生活并非一种免费的生存优势——它需要付出高昂的代价。在食物有限或适婚配偶数量有限的情况下，一群动物会产生危险的竞争。如果这种竞争导致内斗和暴力，那么整个群体最终会浪费宝贵的精力去竞争和打斗。在这种情况下，同样数量的动物最好分开生活。
However, group living is not a freely gained survival benefit—it comes at a high cost. In the presence of food constraints or limited numbers of eligible mates, a herd of animals creates dangerous competition. If this competition leads to infighting and violence, the group ends up wasting precious energy competing and fighting each other. In such a circumstance, the same number of animals would have been better off living separately.
因此，采取群居策略的动物进化出了解决争端的工具，同时将此类争端的能量成本降至最低。这导致了无需实际发生肢体冲突即可发出力量和服从信号机制的发展。鹿和羚羊用角争夺食物和配偶，这是一种比打架便宜得多的竞争形式。熊、猴子和狗则露出牙齿并发出咆哮声以显示攻击性。
Thus, animals who fell into the strategy of group living evolved tools to resolve disputes while minimizing the energetic cost of such disputes. This led to the development of mechanisms to signal strength and submission without having to actually engage in a physical altercation. Deer and antelope lock horns to compete for food and mates, a much cheaper form of competition than fighting. Bears, monkeys, and dogs bare their teeth and growl to show aggression.
这些动物还进化出了表示服从的机制，让它们承认失败，并表明其他人没有必要花费精力伤害它们。狗鞠躬并翻身仰卧；熊则坐下低下头，看向别处；鹿也低下头并压平耳朵。所有这些都提供了一种缓解紧张局势并降低内斗所耗费的精力的机制。
These animals also evolved mechanisms to signal submission, allowing them to acknowledge defeat and show that it is unnecessary for others to spend energy hurting them. Dogs bow and roll over to their backs; bears sit down and look away; and deer lower their heads and flatten their ears. All of this provided a mechanism to ameliorate tensions and lower the amount of energy spent infighting.
凭借表达力量和服从的能力，许多动物能够适应群居生活。大多数哺乳动物谱系都属于以下四种社会系统之一：独居、配偶、妻妾和多雄群体。独居哺乳动物，如驼鹿，成年后大部分时间都是独自生活的，主要是为了交配而聚在一起，如果是雌性，则要养育后代。配偶型哺乳动物，如赤狐和草原田鼠，成对生活在一起，共同养育后代。其他哺乳动物，如骆驼，生活在“闺房”中，即由一头雄性和多头雌性组成的社会群体。还有一些哺乳动物生活在多雄群体中，即由多头雄性和多头雌性共同生活的社会群体。
With the ability to signal strength and submission, many animals were able to make group living work. Most lineages of mammals fell into one of four buckets of social systems: solitary, pair-bonded, harems, and multi-male groups. The solitary mammals, such as moose, spend a large portion of their adult lives on their own, meeting up primarily for mating and, if female, raising children. The pair-bonding mammals, such as red foxes and prairie voles, live together in pairs, raising children together. Other mammals, such as camels, live in harems, which are social groups with a single dominant male and many females. And then there are mammals that live in multi-male groups, social groups with many males and many females living together.
虽然独居和成对的哺乳动物避免了大型社会群体的弊端，但它们也错过了好处。另一方面，后宫和多雄群体享受了大群体的好处，但却要付出竞争的代价。除了表现出侵略性和顺从之外，后宫和多雄群体减少竞争的另一种方式是通过等级制度的僵化。在后宫中，只有一个占主导地位的雄性负责所有的交配；群体中唯一允许的其他雄性是他自己的孩子。
While solitary and pair-bonded mammals avoid the downsides of large social groups, they also miss out on the benefits. On the other hand, harems and multi-male groups reap the benefits of larger groups, but incur the costs of competition. In addition to displays of aggression and submission, another way harems and multi-male groups minimize competition is through hierarchical rigidity. In harems, there is a single dominant male who does all of the mating; the only other males allowed in the group are his own children.
The four common social structures found in mammals
孤 SOLITARY |
配对 PAIR-BONDING |
后宫 HAREMS |
多名男性组成的团体 MULTI-MALE GROUPS |
独立的 Independent |
一男一女 One male, one female |
一名男性，多名女性 One male, many females |
很多男性，很多女性 Many males, many females |
成年后大部分时间独立生活 Live mostly independently in adulthood |
一男一女一起生活并抚养孩子。 One male and one female live and raise children together. |
一只占主导地位的雄性与一群有自己等级制度的雌性生活在一起。 A single dominant male living with a group of females who have their own hierarchy. |
男女等级分开 Separate male and female hierarchy |
老虎 Tigers 捷豹 Jaguar 驼鹿 Moose |
红狐狸 Red foxes 草原田鼠 Prairie voles 巨型水獭 Giant otters 侏儒狨猴 Pygmy marmoset |
蒙古骆驼 Mongolian camels 海豹 Fur seals 大猩猩 Gorillas |
狮子会 Lions 河马 Hippopotamuses 狐猴 Lemurs 黑猩猩 Chimpanzees 狒狒 Baboons 猕猴 Macaque monkeys |
多雄群体也通过严格的等级制度运作：雄性和雌性都有严格的等级制度。低等级的雄性可以加入群体，但它们很少交配，最后才有食物；高等级的雄性首先进食，并负责大部分（如果不是全部的话）交配。
Multi-male groups also work via hierarchical rigidity: There is a strict hierarchy of both males and females. Low-ranking males are allowed in the group, but they do little mating and get the last pick of food; the high-ranking males eat food first and do most, if not all, of the mating.
这些社会群体的等级是如何决定的？很简单——最强壮、最大、最坚韧的人成为主导。角的对峙和牙齿的露出都是为了表明谁会在战斗中获胜，同时避免战斗本身。
How is the hierarchy decided in these social groups? It’s simple—the strongest, biggest, and toughest become dominant. The locking of horns and baring of teeth are all designed to demonstrate who would win in a fight while avoiding the fight itself.
早期灵长类动物也做出了同样的进化权衡——它们进化出了群居生活，为了更好的捕食者的利益而接受攻击的风险避免大群体。灵长类动物遭遇的捕食风险越大，社会群体越大他们创造了回应。像许多现代灵长类动物一样，这些早期灵长类动物可能生活在多雄性群体中，雄性和雌性有等级制度，等级较低的成员得到最差的食物，几乎从未交配过，而等级较高的成员每个人都有自己的选择。它们露出牙齿以示攻击性。我们从化石和对当今许多猴子和猿类社会行为的观察中了解到这一点：黑猩猩、倭黑猩猩和猕猴都以这种方式生活。乍一看，这些早期灵长类动物群体与其他哺乳动物的多雄性群体没有什么不同。但随着研究人员更彻底地观察猴子和猿类的行为，很明显，在社会性方面，灵长类动物与大多数哺乳动物表亲都不一样。早期灵长类动物的社会性发生了一些变化，与早期哺乳动物之前进化出的社会性不同。*
Early primates made this same evolutionary trade-off—they evolved to live in groups, accepting the risk of aggression for the benefit of the better predator avoidance of large groups. The greater the predation risk experienced by primates, the larger the social group they created in response. Like many modern primates, these early primates probably lived in multi-male groups with hierarchies of males and females where lower-ranking members got the worst food and almost never mated and high-ranking members got their pick of each. They bared their teeth as a show of aggression. We know this from fossils and through observing the behavior across many of today’s monkey and ape societies: chimpanzees, bonobos, and macaques all live this way. At first glance, these early primate groups would have looked no different from other mammals’ multi-male groups. But as researchers observed monkey and ape behavior more thoroughly, it became clear that when it comes to sociality, primates are unlike most of their mammalian cousins. Something happened in the sociality of early primates that was different from the sociality that had evolved before in early mammals.*
马基雅维利式的猿猴
Machiavellian Apes
20 世纪 70 年代，灵长类动物学家埃米尔·门泽尔 (Emil Menzel) 对一群黑猩猩进行了实验住在一片一英亩大的森林里。受到托尔曼对老鼠进行心智地图实验的启发，门泽尔开始对黑猩猩进行心智地图实验。他的主要兴趣在于黑猩猩是否能记住隐藏食物的位置。
In the 1970s, the primatologist Emil Menzel was running experiments with a group of chimpanzees living in a one-acre forest. Inspired by Tolman’s experiments on mental maps in rats, Menzel was experimenting with the mental maps of chimpanzees. His main interest was in whether chimpanzees could remember the locations of hidden morsels of food.
门泽尔会将食物藏在这一英亩区域内的任意位置，可能是岩石下或灌木丛中，然后向其中一只黑猩猩透露食物的位置。然后，他会反复将食物放回这些位置。黑猩猩和老鼠一样，能够记住这些确切的位置，学会重新检查这些特定位置以寻找门泽尔藏的食物。但门泽尔开始发现一种与老鼠截然不同的行为，他从未打算研究这种行为，也从未期望发现这种行为。事实上，在仅仅研究空间记忆时，门泽尔就发现了一种令人毛骨悚然的马基雅维利主义行为。
Menzel would hide some food in a random location within this one-acre area, perhaps under a rock or within a bush, and then reveal its location to one of the chimpanzees. He would then place food back in these locations on a recurring basis. Chimpanzees, like rats, were eminently able to remember these exact locations, learning to recheck these specific spots to forage Menzel’s hidden food. But Menzel began spotting behavior that was quite unlike that of a rat, behavior that he had never intended to investigate and had never expected to find. Indeed, while merely investigating spatial memory, Menzel unearthed behavior that was eerily Machiavellian.
当门泽尔第一次向一只名叫贝尔的下级黑猩猩透露藏食物的位置时，她高兴地提醒了其他黑猩猩，并分享了食物。但当占主导地位的雄性洛克过来享用食物时，他把所有的食物都占为己有。洛克这样做了几次之后，贝尔就不再分享食物，开始更多地参与他们使用复杂的策略向 Rock 隐瞒有关食物隐藏位置的信息。
When Menzel first revealed the location of hidden food to one of the subordinate chimpanzees, named Belle, she happily alerted the rest of the group and shared the food. But when the dominate male, Rock, came over to indulge in the treat, he took all the food for himself. After Rock did this a few times, Belle stopped sharing, and began engaging in ever more sophisticated strategies to withhold information about the hidden locations of food from Rock.
起初，贝儿只是坐在食物的秘密位置上，把食物藏起来不让洛克发现，只有当洛克走远时，她才会发现食物并公开吃掉。但当洛克意识到她把食物藏在身下时，他开始逼着她去拿食物。为此，贝儿想出了一个新策略——一旦她看到食物的新位置，她不会立即去找。她会等洛克把目光移开，然后她才会跑向食物。为了应对这一新策略，洛克开始试图欺骗贝儿：他会把目光移开，表现出不感兴趣的样子，一旦贝儿去找食物，他就会转身跑向食物。贝儿甚至开始试图把洛克引向错误的方向，洛克最终意识到了这一欺骗行为，因此，作为回应，他开始朝着与贝儿试图引导他相反的方向寻找食物。
At first, Belle simply sat on top of the secret location of the food to hide it from Rock, and only when he was far away would she uncover and openly eat the food. But when Rock realized she was hiding the food underneath her, he began pushing her to get at the food. In response to this, Belle came up with a new strategy—once she was shown a new location for the hidden food, she didn’t go to it immediately. She would wait for Rock to look away, and then she would run to the food. In response to this new strategy, Rock began trying to trick Belle: He would look away and act uninterested, and once Belle went for the food, he would turn around and run toward it. Belle even began trying to lead Rock in the wrong directions, a deception Rock eventually realized and thus, in response, began to search for food in the opposite direction that Belle would try to lead him.
这种不断升级的欺骗和反欺骗过程表明，洛克和贝儿都能够理解对方的意图（“贝儿试图把我从食物旁引开”，“洛克试图通过看别处来欺骗我”），也明白操纵对方的信念是可能的（“我可以通过假装不感兴趣让贝儿认为我没有在看”，“我可以通过引导洛克朝那个方向让洛克认为食物放在错误的位置”）。自门泽尔的研究以来，许多其他实验也同样发现，猿类实际上可以理解他人的意图。请考虑以下研究：对猿类进行了测试，测试内容是它们区分以下两种行为的能力：“意外”和“故意”行为。向黑猩猩或猩猩展示三个盒子，其中一个盒子里有食物。装满食物的盒子可以与其他盒子区分开来，因为上面有一支笔的粗标记。重复几次，直到它们学会食物总是在有标记的盒子里。然后让一个实验者带着三个盒子进来，通过倾身做标记来故意标记一个盒子，并通过被动地将记号笔放在另一个盒子上来“意外”标记另一个盒子。当猿类被允许来到盒子前寻找食物时，它们会去哪个盒子？它们立即走向实验者“故意”标记的盒子，而忽略“意外”标记的盒子。猿类已经推断出实验者的意图。
This process of ever-escalating deceptions and counter-deceptions reveals that both Rock and Belle were able to understand the other’s intent (“Belle is trying to lead me away from the food,” “Rock is trying to trick me by looking away”), as well as understand that it is possible to manipulate the other’s beliefs (“I can make Belle think I am not looking by pretending to be disinterested,” “I can make Rock think the food is in the wrong location by leading him in that direction”). Since Menzel’s work, numerous other experiments have similarly found that apes can, in fact, understand the intentions of others. Consider the following study: Apes were tested on their ability to tell the difference between “accidental” and “intentional” actions. Show chimpanzees or orangutans three boxes, one of which has food in it. The box filled with food can be differentiated from the others because it has a thick mark from a pen on it. Do this a few times until they learn that food is always in the box with the mark on it. Then have an experimenter come in with the three boxes and intentionally mark one box by leaning over to mark it and “accidently” mark another box by passively dropping the marker on it. When the apes are allowed to come to the boxes to look for food, which box do they go to? They immediately go for the box the experimenter “intentionally” marked and ignore the one that was “accidently” marked. Apes had deduced the intent of the experimenter.
考虑另一项研究。让一只黑猩猩坐在两只实验者身边有食物。一位实验者由于各种原因无法提供食物（有时看不到食物；有时食物卡住了；有时他们表现出丢失食物）。另一位实验者不愿意给他们食物（只是有食物但不给他们）。两位实验者都没有提供食物，但黑猩猩对待这两种情况的方式不同。当有机会在两位实验者之间做出选择时，黑猩猩总是回到似乎无法提供帮助的人身边，并避开那些看起来不愿意的人。黑猩猩似乎能够利用关于另一个人的情况的线索（他们能看到食物吗？他们丢了食物吗？他们只是得不到食物吗？）来推断他们的意图，从而预测这个人将来给他们食物的可能性。
Consider another study. Have a chimpanzee sit across from two experimenters who have food near them. One experimenter is unable to give food for various reasons (sometimes unable to see the food; sometimes the food is stuck; other times they demonstrate losing the food). Another experimenter is unwilling to give them food (simply having food but not giving it). Neither experimenter provided food, and yet chimpanzees treat these two cases differently. When given the chance to pick between these two experimenters, chimps always went back to the person who seemed unable to help and avoided those who seemed unwilling. Chimpanzees seemed able to use cues about the situation of another person (Can they see the food? Did they lose it? Can they just not get it?) to reason about their intentions, and thereby predict the likelihood this other person would give them food in the future.
理解他人的思想不仅需要了解他们的意图，还需要了解他们的知识。贝儿坐在食物上不让洛克发现，这是贝儿操纵洛克知识的尝试。在另一项测试中，黑猩猩有机会玩两副护目镜——一副是透明的，容易看透，另一副是不透明的，很难看透。当黑猩猩有机会向戴着同样护目镜的人类实验者索要食物时，它们知道要去找戴着透明护目镜的人——它们能分辨出戴着不透明护目镜的人护目镜看不到它们。
Understanding the minds of others requires understanding not only their intentions but also their knowledge. Belle sitting on the food to hide it from Rock was Belle’s attempt to manipulate Rock’s knowledge. In another test, chimps were given the chance to play with two sets of goggles—one was transparent and easy to see through, and the other was opaque and hard to see through. When given the chance to ask for food from human experimenters wearing those same goggles, chimps knew to go to the human that was wearing the transparent goggles—they could tell the human wearing the opaque goggles wouldn’t see them.
动物能够推断他人意图和知识的程度在动物心理学中一直存在争议。虽然有证据表明许多灵长类动物（尤其是猿类）具有这种能力，但其他动物的证据尚不明确。其他聪明的动物，像一些鸟类、海豚和狗一样，也能做到这一点。我的观点是，不仅灵长类动物能做到这一点，而且这种能力在早期哺乳动物中并不存在，在人类谱系中，这种壮举出现在早期灵长类动物中（或至少是早期猿类）。即使是社交智力和对人类的关注度很高的狗，也可能无法理解人类可以拥有不同的知识。让狗看到训练员把零食放在一个地方，然后让狗看到别人把零食放在另一个地方（训练员不在场，因此不知道还有另一种零食）。当训练员回来命令“拿零食”时，狗同样有可能跑到任何一个地方，无法识别训练员所指的位置基于训练师知道的地点。
The degree to which animals can infer the intent and knowledge of others continues to be controversial in animal psychology. While there is meaningful evidence that many primates (especially apes) have this ability, the evidence in other animals is less clear. It is possible that other intelligent animals, like some birds, dolphins, and dogs, also can do this. My argument is not that only primates can do this but that this ability was not present in early mammals, and in the human lineage this feat emerged with early primates (or at least by early apes). Even dogs, as socially intelligent and attentive to humans as they are, may be unable to understand that humans can hold different knowledge. Allow a dog to see its trainer put a treat in one location and then allow the dog to see someone else put a treat in a different location (while the trainer is not present and hence unaware of this other treat). When the trainer comes back and commands, “Get the treat,” dogs are equally likely to run to either location, failing to identify the location the trainer is referring to based on which location the trainer knows about.
这种推断某人意图和知识的行为被称为“心智理论”——之所以这样命名，是因为它要求我们有一个关于他人思想的理论。有证据表明，这是一种认知壮举，它出现在早期灵长类动物身上。正如我们将看到的，心智理论可能解释了为什么灵长类动物有如此大的大脑，以及为什么它们的大脑大小与群体规模相关。
This act of inferring someone’s intent and knowledge is called “theory of mind”—so named because it requires us to have a theory about the minds of others. It is a cognitive feat that evidence suggests emerged in early primates. And as we will see, theory of mind might explain why primates have such big brains and why their brain size correlates with group size.
灵长类政治
Primate Politics
非人类灵长类动物最明显的社会行为是梳理毛发——一对猴子会轮流从对方的背上挑出它们自己够不到的污垢和螨虫。在二十世纪上半叶，人们认为这种行为主要是为了卫生。但现在无可争议的是，这种梳理毛发的行为更多的是出于社会目的，而不是卫生目的。梳理毛发所花的时间与身体大小之间没有相关性（如果梳理毛发的功能是为了清洁身体，那么这是可以预料的），但梳理毛发所花的时间与身体大小之间有很强的相关性。梳理和群体大小。此外，不经常被其他动物梳理的个体无法弥补这一点通过更多地梳理自己。每只猴子都有非常特定的梳理伙伴，这些伙伴会持续很长一段时间，甚至一生。
The most obvious social behavior of nonhuman primates is grooming—a pair of monkeys will take turns picking dirt and mites from each other’s backs where they can’t reach themselves. In the first half of the twentieth century, this behavior was believed to be primarily for hygienic purposes. But it is now undisputed that this grooming behavior serves more of a social purpose than a hygienic purpose. There is no correlation between time spent grooming and body size (which you would expect if the function of grooming was for cleaning the body), but there is a strong correlation between time spent grooming and group size. Further, individuals that don’t get groomed by others a lot don’t make up for it by grooming themselves more. And individual monkeys have very specific grooming partners that persist over long periods, even lifetimes.
灵长类动物群体并非由随机互动的个体组成的社会混合体；这些由 15 到 50 只灵长类动物组成的小型社会是由动态和特定关系的子网络组成的。猴子会跟踪并记住群体中的每个个体，并能够识别它们通过外表和声音来识别。它们不仅追踪个体，还追踪个体之间的具体关系。当听到远处一个孩子的求救声时，群体中的个体立即不再看向求救声的方向，而是看向求救孩子的母亲——哦不，爱丽丝会怎么做来帮助她的女儿？或者我们能相信这个孩子吗？让我们看看母亲做什么。
Primate groups are not made up of a social mash of randomly interacting individuals; these mini-societies of fifteen to fifty primates are made up of subnetworks of dynamic and specific relationships. Monkeys keep track of and remember each individual in their group and are able to recognize them by appearance and voice. They keep track not only of individuals but also of the specific relationships between individuals. When a faraway distress call is heard from a child, individuals of the group all immediately look not in the direction of the distress call but at the mother of the distressed child—Oh no, what will Alice do to help her daughter? Or Can we trust this kid? Let’s see what the mother does.
个体之间的关系不仅是家庭关系，也是等级关系。长尾黑尾猴有一种接近-撤退的习惯来表示统治和服从；当一个地位较高的个体走近时向地位较低的个体屈服，地位较低的个体就会退却。这些支配关系在各种情况下都存在：当猴子 A 在一种情况下对 B 表示服从时，几乎总是 A 也会在另一种情况下服从 B。这些支配关系是可传递的：如果你看到猴子 A 服从 B，而 B 服从 C，那么几乎肯定是A 服从 C。这些等级制度通常会持续很多年，甚至几代人。这些支配和服从的信号并不是一次性的展示；它们代表着明确的社会等级制度。
The relationships between individuals are not only familial but also hierarchical. Vervet monkeys have an approach-retreat routine to signal dominance and submission; when a higher-ranking individual walks toward a lower-ranking individual, the lower-ranking individual retreats. These dominance relationships persist across contexts: When monkey A expresses submission to B in one situation, it is almost always the case that A will also submit to B in another situation. These dominance relationships are transitive: If you see monkey A submit to B, and B submit to C, then it is almost definitely the case that A will submit to C. And these hierarchies often persist for many years, even generations. These signals of dominance and submission are not one-off displays; they represent an explicit social hierarchy.
灵长类动物对违反社会等级的互动极为敏感。在 2003 年进行的一项研究中，实验人员获取了一群狒狒中不同成员发出支配或服从声音的录音，然后在狒狒附近设置扬声器，以便它们能够播放这些录音。当他们播放一只等级较高的狒狒发出统治声音，然后播放一只等级较低的狒狒发出屈服声音的录音时，没有狒狒看向扬声器——这在确立对等级较低的狒狒的统治地位方面并不奇怪。然而，当他们播放任何等级较低的狒狒发出统治声音，然后播放一只等级较高的狒狒发出屈服声音的录音时——这违反了等级制度——狒狒就会惊慌失措，盯着扬声器看到底发生了什么。就像当一个书呆子打了那个受欢迎的恶霸的脸时，全班同学都忍不住目瞪口呆：这真的发生了吗？
Primates are extremely sensitive to interactions that violate the social hierarchy. In a study done in 2003, experimenters obtained audio recordings of different members of a group of baboons making dominance or submission noises, and then set up speakers near the baboons so they could play back these recordings. When they played a recording of a higher-ranking baboon making a dominance noise followed by a recording of a lower-ranking baboon making a submission noise, no baboons looked over to the speakers—nothing surprising about someone establishing their dominance over someone ranked below them. However, when they played a recording of any lower-ranking baboon making a dominance noise followed by a recording of a baboon ranked above them making a submission noise—a violation of the hierarchy—baboons freaked out and stared at the speakers to see what in the world had just happened. Like when a nerd smacks the popular bully in the face, the class can’t help but gawk: Did that really just happen?
这些猴子社会的独特之处不在于社会等级制度的存在（许多动物群体都有社会等级制度），而在于等级制度的构建方式。如果你研究不同猴子群体的社会等级制度，你会注意到，位于顶端的往往不是最强壮、最大或最具攻击性的猴子。与大多数其他社会动物不同，对于灵长类动物来说，不仅是体力决定了一个人的社会地位，还有政治权力。
What makes these monkey societies unique is not the presence of a social hierarchy (many animal groups have social hierarchies), but how the hierarchy is constructed. If you examined the social hierarchy of different monkey groups, you would notice that it often isn’t the strongest, biggest, or most aggressive monkey who sits at the top. Unlike most other social animals, for primates, it is not only physical power that determines one’s social ranking but also political power.
就像许多早期人类文明一样（不幸的是，今天仍然有很多），决定猴子在群体中地位的一件事是它出生的家庭。在灵长类社会群体中，家庭往往有等级制度。以下是雌性等级制度的常见结构：等级制度的顶端是地位最高的家庭中最年长的女性，其次是她的后代，然后是地位第二高的家庭中最年长的成员，然后是她的后代，依此类推等等。当女儿的母亲去世时，她往往会继承她母亲的军衔。
As was the case in many early human civilizations (and unfortunately still many today), one thing that determines a monkey’s place in its group is the family it is born into. In primate social groups, there tends to be a hierarchy of families. Here is a common structure for female hierarchies: At the top of the hierarchy is the eldest female of the highest-ranking family, followed by her offspring, then the eldest member of the second-highest-ranking family, followed by her offspring, and so on and so forth. And when a daughter’s mother dies, she tends to inherit the rank of her mother.
与非灵长类动物通常将力量与等级联系起来的做法截然不同的是，一个强大家族中虚弱的幼年成员很容易吓跑一个比她地位低的家族中体型更大、更强壮的成年猴子。事实上，孩子自己很清楚自己在社会结构中的地位——即使是年幼的孩子也会经常挑战地位较低的家族中的成年人，但他们不会挑战成年人来自较高等级的家族。
In a clear departure from the typical nonprimate association of strength with rank, a weak and frail juvenile member of a powerful family can easily scare off a much larger and stronger adult monkey from a family that ranks below her. In fact, the child herself is clearly aware of her place in the social structure—even young children will regularly challenge adults of lower-ranking families, but they won’t challenge adults of higher-ranking families.
人类社会也是如此——王朝权力斗争无休无止，家族兴衰更替——猴子王朝也同样兴衰更替。家族面临着巨大的压力，需要提高自己的地位。地位较高的猴子可以挑选食物、梳理伴侣、配偶和休息地点。一个人的进化适应性随着地位的提高而提高；地位较高的猴子有更多的孩子，而且不太可能死于疾病。因此，如果一个高等级家族的人数减少到一定程度，一个低等级家族就会发动协同叛乱；低等级家族会不断发起攻击，直到高等级家族屈服，此时一个新的层次结构已经建立。
And as it goes with human societies—with endless oscillations of dynastic power struggles of families rising and collapsing—monkey dynasties rise and fall too. Families face incredible pressure to improve their position. Higher-ranking monkeys get their pick of food, grooming partners, mates, and resting sites. One’s evolutionary fitness improves with one’s rank; higher-ranking monkeys have more children and are less likely to die from disease. And so if a high-ranking family sufficiently dwindles in number, a lower-ranking family will wage a coordinated mutiny; the lower-ranking family will make persistent aggressive challenges until the higher-ranking family submits, at which point a new hierarchy has been established.
此类叛乱并非不可避免；人数较少的高地位家族可以与家族外的成员结盟，以帮助巩固其地位。事实上，在出现攻击性行为时，大约有 20% 的时间，附近的其他猴子会通过与攻击者或防御者联手做出回应。大多数时候是家庭成员前来帮忙，但大约三分之一的时间是来的非家庭成员。建立这种联盟的能力似乎是决定个体等级的主要因素之一：等级较高的猴子往往更善于从不相关的个体中招募盟友，等级逆转最常发生在猴子失败的时候来招募这样的盟友。
Such mutinies are not inevitable; high-ranking families that are few in number can form alliances with members outside of their families to help solidify their position. Indeed, about 20 percent of the time that an aggressive display is made, other nearby monkeys will respond by joining forces with either the attacker or the defender. Most of the time it is family members that come to help, but about a third of the time it is nonfamily members that come. It seems that the ability to forge such allyships is one of the primary determinants of an individual’s rank: higher-ranking monkeys tend to be better at recruiting allies from unrelated individuals, and hierarchy reversals most often happen when monkeys fail to recruit such allies.
猴子政治发生在这些联盟关系的动态中，这些联盟关系不是通过固定的家庭关系形成的，而是在冲突中培养和支持他人。联盟和培养伙伴关系代表一种常见的关系，我们称之为友谊：猴子最常拯救那些它们以前建立了梳理伙伴关系。猴子可以通过善举获得回报，即使对那些目前不是朋友的人也是如此。如果猴子 A 不遗余力地为猴子 B 梳理毛发，那么下次 A 遇到 B 时，B 更有可能跑去保护 A发出“救救我”的声音。在冲突中竭尽全力支持他人也是如此——猴子往往会跑去保护那些逃跑的人。自我防御。在这些同盟关系中也存在着信任的概念：当一只黑猩猩面临两个选择：（a）实验者直接给她一份普通的零食，或（b）实验者给另一只黑猩猩一份美味的零食，希望另一只黑猩猩能分享，只有当另一只黑猩猩是梳理伙伴时，黑猩猩才会选择（b）。否则，它们会选择更糟糕的只为他们自己准备的食物。
Monkey politics happens in the dynamics of these allyships, which are forged not through fixed familial relationships but grooming and supporting others in conflicts. Allyships and grooming partnerships represent a common relationship, what we would call a friendship: monkeys most often rescue those whom they have previously formed grooming partnerships with. And a monkey can earn reciprocity through acts of kindness even to those they are not currently friends with. If monkey A goes out of its way to groom monkey B, then B is substantially more likely to run to A’s defense next time A makes a “help me” vocalization. This is also true with going out of one’s way to support others in conflicts—monkeys tend to run to the defense of those who have run to their own defense. In these allyships is also a notion of trust: When a chimpanzee is given the option of (a) an experimenter giving her a mediocre snack directly, or (b) an experimenter giving another chimp, who will hopefully share, an amazing snack, chimps chose (b) only if the other chimp is a grooming partner. Otherwise they took the worse food just for themselves.
这些盟友关系对猴子的政治地位和生活质量有很大影响。那些拥有权力的人从与低等级盟友建立足够的联盟中受益，而低等级的猴子可以通过与合适的高等级家庭建立友谊来大大改善他们的生活。拥有强大梳理伙伴的低等级个体受到的骚扰要少得多，即使高等级盟友不在视线范围内；群体中的每个人都知道不要惹詹姆斯，除非你想来对付基思。高等级的猴子对与它们结成同盟的低等级个体更加宽容，这让它们获取更多食物。
These allyships have a big impact on a monkey’s political standing and quality of life. Those with power benefit from forging a sufficient coalition of lower-ranking allies, and low-ranking monkeys can substantially improve their lives by forging friendships with the right high-ranking families. Low-ranking individuals with powerful grooming partners get harassed much less, even when the high-ranking ally is out of sight; everyone in the group knows Don’t mess with James unless you want to deal with Keith. High-ranking monkeys are more tolerant of the low-ranking individuals with whom they have forged allyships, giving them more access to food.
猴子的许多社会行为都表明了它们具有令人难以置信的政治远见。猴子更喜欢与那些排名高于自己的同类。猴子喜欢与排名更高的同类交配。猴子们争相梳理毛发与高等级个体的合作。当发生争端时，猴子倾向于与地位较高的个体。地位较高的母亲的孩子是最受欢迎的玩伴。
Much of monkey social behavior suggests an incredible degree of political forethought. Monkeys prefer to invest in relationships with those ranked higher than themselves. Monkeys prefer to mate with higher-ranking members of the group. Monkeys compete to groom with high-ranking individuals. When a dispute breaks out, monkeys are biased toward joining forces with the higher-ranking individual. The children of high-ranking mothers are the most popular playmates.
高等级的猴子还表现出聪明才智，它们会选择与低等级的成员交朋友。在一项研究中，不同的低等级猴子被训练完成特定的任务来获取食物，高等级的猴子很快就和那些拥有特殊技能的猴子交上了朋友，即使在他们分手后，它们仍坚持这种梳理关系。没有立即获得食物的前景：我可以看出你很有用；让我带我将为你提供庇护。
High-ranking monkeys also exhibit a cleverness in which lower-ranking members they choose to befriend. In a study where different low-ranking monkeys were trained to do specific tasks to obtain food, high-ranking monkeys quickly befriended those who had specialized skills, and they persisted in these grooming partnerships even after there was no immediate prospect of gaining food: I can see you are useful; let me take you under my wing.
猴子在冲突后也表现出政治智慧。在激烈的互动之后，他们会不遗余力地试图“和解”，尤其是那些与非家庭成员相处。他们经常试图拥抱和打架的人，并试图与这些人的家人和好，与他们打架的人的家人在一起的时间是平时的两倍最近吵架了。
Monkeys also show political cleverness in the aftermath of conflict. They go out of their way to try and “make up” after aggressive interactions, especially those with nonfamily members. They often seek to hug and groom those they fought with, and they also try to make up with the individuals’ families, spending double the usual amount of time with family members of those they have recently quarreled with.
政治智慧的军备竞赛
The Arms Race for Political Savvy
早期灵长类动物的进化轨迹不知何故导致了现代灵长类动物中出现的一系列极为复杂的社会行为的发展。在这些行为中，我们看到了人类相互交流的行为基础的线索。灵长类动物为何会进化出这些本能尚不清楚，但这可能与二叠纪-三叠纪灭绝事件后早期灵长类动物所处的独特环境有关。
Somehow the evolutionary trajectory of early primates led to the development of the incredibly broad suite of complex social behaviors that are seen across modern species of primates. And within these behaviors we see hints of the behavioral foundation of how humans tend to interact with each other. Why primates evolved these instincts is not exactly clear, but it may have had to do with the unique niche early primates found themselves in in the aftermath of the Permian-Triassic extinction event.
早期灵长类动物似乎有一种独特的饮食习惯，即直接在树梢上采摘果实——它们是食果动物。它们在果实成熟但未落到森林地面之前就从树上采摘果实。这使得灵长类动物能够轻松获取食物，而不必面对来自其他物种的太多竞争。这种独特的生态位可能为早期灵长类动物提供了两份礼物，为它们独特的大脑和复杂的社会群体打开了大门。首先，轻松获取水果为早期灵长类动物提供了充足的卡路里，为将能量花在更大的大脑上提供了进化选择。其次，也许更重要的是，它为早期灵长类动物提供了充足的时间。
Early primates seemed to have had a unique diet of foraging fruit directly in treetops—they were frugivores. They plucked fruit from trees right after it ripened but before it fell to the forest floor. This allowed primates to have easy access to food without much competition from other species. This unique ecological niche may have offered early primates two gifts that opened the door to their uniquely large brains and complex social groups. First, easy access to fruit gave early primates an abundance of calories, providing the evolutionary option to spend energy on bigger brains. And second, and perhaps more important, it gave early primates an abundance of time.
在动物界中，空闲时间极其稀少；大多数动物别无选择，只能用以下方式填满它们每天的每一刻：进食、休息和交配。但是这些以水果为食的灵长类动物不需要像其他动物那样花费大量时间觅食，因此在寻求攀登社会等级时，这些灵长类动物有了新的进化选择：它们不必花费精力进化出更大的肌肉来战斗，爬上社会的顶峰，而是可以花费精力进化出更大的大脑，通过政治手段爬上社会的顶峰。
Free time is extremely rare in the animal kingdom; most animals have no choice but to fill every moment of their daily calendar with eating, resting, and mating. But these frugivorous primates didn’t have to spend nearly as much time foraging as other animals did, so when seeking to climb the social hierarchy, these primates had a new evolutionary option: instead of spending energy evolving bigger muscles to fight their way to the top, they could spend energy evolving bigger brains to politick their way to the top.
因此，灵长类动物似乎把他们的空闲时间都用来参加政治活动了。如今的灵长类动物每天有20% 的时间用于社交，这比以前多得多。比大多数其他哺乳动物都要多。研究表明，这种社交时间与灵长类动物的空闲时间长短有因果关系；随着空闲时间的增加（通过提供更容易获取食物的机会），灵长类动物花更多时间社交。
So primates seemed to have filled their open calendars with politicking. Today’s primates spend up to 20 percent of their day socializing, a much larger amount of time than most other mammals. And it has been shown that this social time is causally related to how much free time primates have; as more free time is given (by providing easier access to food), primates spend more time socializing.
这引发了一场全新的进化军备竞赛：一场政治智慧之战。任何天生就懂得如何讨好和争取盟友的灵长类动物都能更好地生存下来并生育更多后代。这给其他灵长类动物带来了更大的压力，迫使它们进化出更聪明的政治机制。事实上，灵长类动物的新皮层大小不仅与社会群体大小有关，还与社交能力。这场军备竞赛的产物似乎是人类许多社交本能的蓬勃发展，既有好的（友谊、互惠、和解、信任、分享），也有坏的（部落主义、裙带关系、欺骗）。虽然这些行为变化的许多方面不需要任何特别聪明的新大脑系统，但这种政治活动背后确实存在着一种智力壮举：参与心理理论的能力。
This created a completely new evolutionary arms race: a battle for political savvy. Any primate born with better tricks for currying favor and gaining allies would survive better and have more babies. This put more pressure on other primates to evolve smarter mechanisms for politicking. Indeed, neocortex size of primates is correlated not only with social-group size but also with social savviness. An outgrowth of this arms race seems to have been a blossoming of many human social instincts, both the good (friendships, reciprocity, reconciliation, trust, sharing) and the bad (tribalism, nepotism, deception). While many aspects of these behavioral changes did not require any particularly clever new brain systems, there was indeed an intellectual feat underlying this politicking: the ability to engage in theory of mind.
如果一个物种连最基本的、最原始的心理理论都没有，那么政治智慧如何可能就不得而知了——只有通过这种能力，个体才能推断出其他人想要什么，从而弄清楚该和谁亲近，以及如何亲近。只有通过心理理论，灵长类动物个体才能知道不要惹一个有高级别朋友的低级别个体；这需要了解高级别个体的意图以及他们在未来情况下会做什么。只有通过这种心理理论能力，你才能弄清楚谁可能在未来变得强大，你需要和谁交朋友，你可以欺骗谁。
It isn’t clear how political savviness would even be possible if a species did not have at least a basic and primitive version of theory of mind—only through this ability can individuals infer what others want and thereby figure out whom to cozy up to and how. Only through theory of mind can individual primates know not to mess with a low-ranking individual with high-ranking friends; this requires understanding the intent of the high-ranking individuals and what they will do in future situations. Only through this ability of theory of mind can you figure out who is likely to become powerful in the future, whom you need to make friends with, and whom you can deceive.
因此，这或许就是为什么灵长类动物的大脑会长得这么大，为什么它们的大脑大小与社会群体大小相关，以及为什么灵长类动物会进化出推理他人思想的能力。当然，问题是，灵长类动物的大脑是如何做到这一点的？
So this may be why primates began growing such big brains, why their brain size is correlated with social group size, and why primates evolved the ability to reason about the minds of others. The question is, of course, how do primate brains do this?
16
16
如何模仿他人的思维
How to Model Other Minds
七千万年前，我们的哺乳动物祖先的大脑重量不到半克。到一千万年前，我们的猿类祖先出现时，大脑已经扩展到大约 350 克。这几乎是大脑体积的 1000 倍。如此大的扩张对跨时间（早期哺乳动物到早期灵长类动物）和跨物种（当今小鼠大脑到当今黑猩猩大脑）的大脑区域关联提出了挑战。哪些大脑区域是真正新的，哪些只是同一大脑区域的扩展版本？
OUR MAMMALIAN ANCESTOR from seventy million years ago had a brain that weighed less than half a gram. By the time our ape ancestors arrived ten million years ago, it had expanded to about three hundred fifty grams. This is almost a thousandfold increase in brain size. Such a large expansion presents a challenge in relating brain areas across time (early mammals to early primates) and across species at a point in time (today’s mouse brain to today’s chimpanzee brain). Which brain areas are truly new and which are merely expanded versions of the same thing?
显然，大脑中的一些结构会随着身体尺寸的增大而自然增大，而不会对其功能产生任何重大变化。例如，更大的身体意味着更多的触觉和痛觉神经，这意味着有更多的新皮层空间来处理这些感觉信号。早期猿类的体感皮层的表面积显然比早期哺乳动物的大得多，尽管它执行相同的功能。更大的眼睛和肌肉以及任何其他需要输入或输出神经的东西也是如此。
Clearly some structures in brains will scale naturally with body size without any meaningful change to their function. For example, a bigger body means more touch and pain nerves, which means more neocortical space for processing these sensory signals. The surface area of the somatosensory cortex of the early apes was obviously much bigger than that of early mammals even though it performed the same function. Same for bigger eyes and muscles and anything else that requires incoming or outgoing nerves.
此外，可以在结构中添加更多神经元来改善其性能，而无需从根本上改变其功能。例如，如果基底神经节大 100 倍，它可能能够实现更多动作和奖励之间的关联，同时仍然执行相同的功能：实现时间差分学习算法。同样，灵长类动物的视觉皮层比啮齿类动物的视觉皮层大得多，即使考虑到大脑的缩放比例。毫不奇怪，灵长类动物在视觉处理的许多方面都比啮齿类动物更好。但是大脑皮层的视觉区域在灵长类动物中并不发挥某些独特的功能；灵长类动物只是按比例为相同功能分配了更多的空间，从而获得了更好的表现。
Further, more neurons can be added to a structure to improve its performance without fundamentally changing its function. For example, if the basal ganglia were one hundred times larger, it might enable associations between many more actions and rewards while still fundamentally performing the same function: implementing a temporal difference learning algorithm. Similarly, the visual cortex of primates is massively larger than that of rodents, even accounting for brain scaling. Unsurprisingly, primates are better than rodents at many aspects of visual processing. But the visual area of the neocortex does not perform some unique function in primates; primates simply dedicated more space, proportionally, to the same function and got better performance.
然后还有模糊区域——这些结构非常相似但略有修改，在新旧边缘摇摆不定。一个例子是大脑皮层中新的感官处理层次。灵长类动物的视觉新皮层有许多层次区域，处理从一个区域跳到另一个区域。这些区域仍然处理视觉输入，但新层次的增加使它们在质量上有所不同。一些区域对简单的形状做出反应；其他区域对面部做出反应。
And then there are the fuzzy areas—those structures that are very similar but slightly modified, teetering on the edge of new and old. An example of this is new hierarchical layers of sensory processing in the neocortex. Primates have many hierarchical regions of the visual neocortex with processing hopping from one region to the next. These areas still process visual input, but the addition of new hierarchical layers makes them qualitatively different. Some areas respond to simple shapes; other areas respond to faces.
但当然，也存在真正新的大脑区域——具有完全独特连接性并执行新功能的结构。
But there are also, of course, truly new brain regions—structures with completely unique connectivity that perform novel functions.
所以问题是：在早期灵长类动物的大脑中，有多少只是简单地扩大了（无论是按比例还是不按比例），又有多少是新的？大多数证据表明，尽管尺寸急剧扩大，但我们的灵长类祖先和当今灵长类动物的大脑与早期哺乳动物的大脑大致相同。更大的后脑、更大的基底神经节、更大的新皮层，但所有相同的区域仍然连接在一起。基本方式相同。这些早期灵长类动物确实将更多的新皮层用于某些功能，例如视觉和触觉，但功能和连接性仍然大致相同。添加了新的层次结构，感官信息从一个新皮层跳跃到另一个新皮层，从而能够逐渐形成更抽象的表征。但这主要只是为了提高表现。
So the question is: In the brain of early primates, how much was simply scaled up (whether proportionately or disproportionately) and how much was new? Most evidence suggests that despite the dramatic expansion in size, the brain of our primate ancestor, and of primates today, was largely the same as that of the early mammals. A bigger hindbrain, a bigger basal ganglia, a bigger neocortex, but still all the same regions connected in all the same fundamental ways. These early primates indeed dedicated disproportionately more neocortex to certain functions, such as vision and touch, but still, the functions and connectivity were largely the same. New hierarchical layers were added, with sensory information hopping from one neocortical layer to another, enabling progressively more abstract representations to be formed. But this was mostly just performance-enhancing.
那么，灵长类动物的惊人智慧，以及所有他们的心智理论、政治手段和诡计，难道仅仅是脑力扩展的结果吗？
So might it be the case that the surprising smarts of primates, with all their theory of mind, politicking, and trickery, was a consequence of nothing more than brain scaling?
图 16.1
Figure 16.1
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
早期灵长类动物的新皮层区域
The New Neocortical Regions of Early Primates
虽然早期灵长类动物的大脑大部分只是哺乳动物大脑的放大版，但实际上，新皮质的某些区域确实是新的。我们可以将灵长类谱系中出现的这些新皮质区域分为两类。第一类是颗粒前额皮质 (gPFC)，这是一种新的额叶皮质的补充。*这种较新的颗粒状前额叶皮质包裹着更古老的无颗粒状前额叶皮质 (aPFC)。新皮质的第二个新区域，我称之为灵长类感觉皮质 (PSC)，是灵长类动物中出现的几个新感觉皮质区域的融合。* gPFC 和 PSC 彼此高度互联，构成了它们自己的新额叶网络以及感觉新皮质区域。
Although most of the brain of the early primates was merely a scaled-up mammal brain, there were, in fact, certain areas of neocortex that were truly new. We can categorize these new neocortical areas that emerged within the primate lineage into two groups. The first is the granular prefrontal cortex (gPFC), which was a new addition to the frontal cortex.* This newer granular prefrontal cortex wraps around the much older agranular prefrontal cortex (aPFC). The second new area of neocortex, which I will call the primate sensory cortex (PSC), is an amalgamation of several new areas of sensory cortex that emerged in primates.* The gPFC and PSC are extremely interconnected with each other, making up their own new network of frontal and sensory neocortical regions.
是什么让这些区域变得“新”？不是它们的微电路；所有这些区域仍然是新皮层，并且与哺乳动物新皮层的其他区域具有相同的柱状微电路。它们的输入和输出连接使它们变得新；这些区域构建了一个生成模型，从根本上解锁了新的认知能力。
What makes these areas “new”? It isn’t their microcircuitry; all these areas are still neocortex and have the same general columnar microcircuitry as other areas of neocortex across mammals. It is their input and output connectivity that renders them new; it is what these areas construct a generative model of that unlocked fundamentally new cognitive abilities.
正如我们在突破#3中看到的那样，aPFC 受损的人会出现明显而严重的症状，例如运动不能性缄默症，患者会变得完全沉默且失去意识。
As we saw in breakthrough #3, a human with damage to the aPFC has obvious and severe symptoms, such as akinetic mutism, in which patients become completely mute and intentionless.
与 aPFC 损伤的惊人症状相比，周围颗粒状前额皮质的损伤通常只会导致轻微的症状。事实上，这些区域损伤造成的损害非常小，以至于 20 世纪 40 年代的许多神经科学家怀疑这些区域是否缺乏任何功能意义。当时一个著名的案例是一位名叫 KM 的癫痫患者，为了治疗癫痫，他切除了三分之一的额叶皮层。手术后，KM 似乎没有出现任何智力或感知能力。在切除三分之一的额叶皮层后，KM 的智商没有变化——如果有的话，它反而增加了。用当时一位神经学家的话来说，颗粒前额叶皮层的功能皮层是一个“谜”。
In contrast to the alarming symptoms of aPFC damage, damage to the surrounding granular prefrontal cortex often results in minimal symptoms. In fact, the impairment from damage to these areas is so minimal that many neuroscientists in the 1940s wondered if these areas lacked any functional significance at all. A famous case study at the time was of a seizure patient named K.M. who had a third of his frontal cortex removed to treat seizures. After the surgery, K.M. seemed to have no deficits in intellect or perception whatsoever. K.M.’s IQ after the removal of a third of his frontal cortex was unchanged—if anything, it increased. In the words of a neuroscientist at the time, the function of granular prefrontal cortex was a “riddle.”
图 16.2：哺乳动物共享的新皮层区域和灵长类动物的新区域
Figure 16.2: Shared neocortical regions across mammals and new regions of primates
原创艺术由 Mesa Schumacher 创作
Original art by Mesa Schumacher
塑造自己的思维
Modeling Your Own Mind
考虑一下 2001 年的一项研究。人类受试者被放入 fMRI 机器中，并被展示一系列图片。每看一张图片，受试者都会被问到“这让你感觉如何？”或一些关于图片内容的平凡问题，例如“这张照片是在室内还是室外拍摄的？”这两项任务都激活了 aPFC，这是有道理的，因为它们都需要渲染内部模拟；要么是图片周围的世界，以决定它更可能是在室内还是室外，要么是你自己的想法和感觉。但只有当人类被问到他们对图片的感受时，aPFC才会激活。gPFC 亮了。
Consider a study done in 2001. Human subjects were put in an fMRI machine and shown a series of pictures. With each picture, subjects were asked “How does this make you feel?” or some mundane question about the contents of the image, such as “Was this picture taken indoors or outdoors?” Both tasks activated the aPFC, which makes sense, since both require rendering an inner simulation; either of the world around the picture to decide if it is more likely to be inside or outside, or of your own thoughts and sensations. But it was only when humans were asked how they felt about a picture that the gPFC lit up.
现在有大量实验证实了这一点。颗粒状前额叶皮层在需要自我参照的任务中变得特别活跃，例如评估自己的性格特征、一般的自我相关思维游移、考虑自己的感受、思考自己的意图以及思考关于你自己的一般信息。
There have now been numerous experiments confirming this. The granular prefrontal cortex becomes uniquely active during tasks that require self-reference, such as evaluating your own personality traits, general self-related mind wandering, considering your own feelings, thinking about your own intentions, and thinking about yourself in general.
有了这个线索，颗粒状前额叶皮层只通过自我参照而被激活，我们是否会错过颗粒状前额叶损伤的一些细微但关键的损害？
With this clue the granular prefrontal cortex is uniquely activated by self-reference, might we have missed some subtle but crucial impairments of granular prefrontal damage?
2015 年，科学家进行了以下研究。他们给参与者一个中性提示词（例如，鸟或餐厅），并要求他们向实验者讲述与该词相关的不同故事。其中一些参与者身体健康，一些前额叶皮质颗粒区域受损，一些海马体受损。
In 2015, scientists did the following study. They gave participants a neutral cue word (e.g., bird or restaurant) and asked them to tell the experimenter different narratives of themselves associated with that word. Some of these participants were healthy, some had damage to areas of the granular prefrontal cortex, and some had damage to the hippocampus.
在这些情况下，人们的叙述有何不同？前额叶颗粒区受损但前额叶皮质和海马体完好的人能够想象复杂的场景，细节丰富，但他们无法想象自己身处这样的场景。他们有时甚至会完全忽略自己的叙述。海马体受损似乎产生了相反的效果——患者可以很好地想象自己身处过去或未来的情境，但很难构建世界的外部特征；他们无法详细描述任何周围的元素。
How did people’s narratives differ across these conditions? Humans with damage to the granular prefrontal areas but with the aPFC and hippocampus intact were able to imagine complex scenes, rich with detail, but they were impaired at imagining themselves in such scenes. They sometimes even completely omitted themselves from their narratives. Damage to the hippocampus seemed to have the opposite effect—patients could imagine themselves in past or future situations just fine but struggled to build external features of the world; they were unable to describe in detail any of the surrounding elements.
这表明，粒状前额皮质在你将自己（你的意图、感受、想法、个性和知识）投射到渲染的模拟中的能力中起着关键作用，无论这些模拟是关于过去还是想象的未来。这些模拟是在老鼠的大脑中进行的，老鼠的大脑没有 gPFC，只有 aPFC 和海马体，显示出渲染外部世界的证据，但没有任何迹象表明它们真正将任何有意义的自身模型投射到这些模拟中。
This suggests that the granular prefrontal cortex plays a key role in your ability to project yourself—your intentions, feelings, thoughts, personality, and knowledge—into your rendered simulations, whether they are about the past or some imagined future. The simulations run in rat brains, which has no gPFC and only an aPFC and hippocampus, show evidence of rendering an external world, but there is nothing to suggest that they truly project any meaningful model of themselves into these simulations.
颗粒前额皮质受损也可能影响人们对自己的模型，不仅影响他们在心理模拟中的自我模型，还影响他们在当下的自我模型。一些颗粒前额皮质受损的人会出现镜像符号综合症，这种情况下他们不再能识别镜子中的自己。这些病人坚称他们在镜子中看到的人不是他们。您当前对自己思想的塑造与您在想象中投射的思想似乎密切相关。
Damage to the granular prefrontal cortex may also affect people’s models of themselves not only in their mental simulations, but also in the present. Some people with gPFC damage develop mirror-sign syndrome, a condition in which they no longer recognize themselves in a mirror. These patients insist that the people they see in a mirror are not them. Your modeling of your own mind in the present and the mind you project in your imagination seem to be intimately related.
如果你遵循它们的输入/输出连接，那么灵长类动物的新区域参与塑造你自己的思维的想法是有道理的。较老的哺乳动物 aPFC 直接从杏仁核和海马体，而新的灵长类动物 gPFC 几乎不接收杏仁核或海马体输入或任何直接感官输入。相反，灵长类动物 gPFC 的大部分输入直接来自较旧的 aPFC。
The idea that the new primate areas take part in modeling your own mind makes sense when you follow their input/output connectivity. The older mammalian aPFC gets input directly from the amygdala and hippocampus, while the new primate gPFC receives almost no amygdala or hippocampal input or any direct sensory input at all. Instead, the primate gPFC gets most of its input directly from the older aPFC.
对此的一种解释是，这些新的灵长类动物区域正在构建较老的哺乳动物 aPFC 和感觉皮层本身的生成模型。正如 aPFC 构建杏仁核和海马体活动的解释（发明“意图”）一样，也许 gPFC 构建了 aPFC 意图模型的解释——可能发明了所谓的思想。也许 gPFC 和 PSC 构建了一个自身内部模拟的模型，以根据感觉新皮层中的知识来解释 aPFC 中的意图。
One interpretation of this is that these new primate areas are constructing a generative model of the older mammalian aPFC and sensory cortex itself. Just as aPFC constructs explanations of amygdala and hippocampus activity (invents “intent”), perhaps the gPFC constructs explanations of the aPFC’s model of intent—possibly inventing what one might call a mind. Perhaps the gPFC and PSC construct a model of one’s own inner simulation to explain one’s intentions in the aPFC given knowledge in the sensory neocortex.
让我们用一个思想实验来直观地了解一下这意味着什么。假设你把我们的祖先灵长类动物放在迷宫中。当它到达一个选择点时，它会向左转。假设你可以问它的不同大脑区域为什么动物向左转。你会在每个抽象层次上得到非常不同的答案。条件反射会说，因为我有一个进化上硬编码的规则，要转向来自左边的气味。脊椎动物结构会说，因为向左走可以最大化预测的未来回报。哺乳动物结构会说，因为左边有食物。但灵长类动物结构会说，因为我饿了，饿的时候吃东西感觉很好，据我所知，向左走会有食物。换句话说，gPFC 构建了对模拟本身的解释，对动物想要什么、知道什么和想什么的解释。心理学家和哲学家称之为元认知：思考思考的能力。
Let’s use a thought experiment to build some intuition about what this means. Suppose you put our ancestral primate in a maze. When it reached a choice point, it turned left. Suppose you could ask its different brain areas why the animal turned left. You would get very different answers at each level of abstraction. Reflexes would say, Because I have an evolutionarily hard-coded rule to turn toward the smell coming from the left. Vertebrate structures would say, Because going left maximizes predicted future reward. Mammalian structures would say, Because left leads to food. But primate structures would say, Because I’m hungry, eating feels good when I am hungry, and to the best of my knowledge, going left leads to food. In other words, the gPFC constructs explanations of the simulation itself, of what the animal wants and knows and thinks. Psychologists and philosophers call this metacognition: the ability to think about thinking.
哺乳动物在内部模拟外部世界时发现的东西，在某种意义上，与它们对外部世界的知识是一回事。当哺乳动物模拟沿着一条小路行走，其感觉新皮层呈现的模拟结果显示小路的尽头有水时，这与“知道”小路尽头有水是一样的。虽然感觉新皮层中较老的哺乳动物区域会呈现模拟结果外部世界（包含知识）的大脑皮层中新的灵长类动物区域（我一直称之为灵长类动物感觉皮层）似乎创建了这种知识本身的模型（PSC 区域从感觉大脑皮层的各个区域获得输入）。这些新的灵长类动物区域试图解释为什么感觉大脑皮层认为食物在那里，为什么动物对外部世界的内在模拟是这样的。答案可能是：因为上次我去那里时看到了水，因此当我模拟回到那里时，我会在想象中看到水。换言之：因为我上次在那里看到了水，所以我现在知道水在那里，尽管之前我不知道。
What mammals find in their inner simulations of the external world is, in some sense, the same thing as their knowledge about the external world. When a mammal simulates going down a path and its sensory neocortex renders a simulation that contains water at the end of the path, this is the same thing as “knowing” that water exists at the end of the path. While the older mammalian areas of the sensory neocortex render the simulation of the external world (containing knowledge), the new primate areas of neocortex (what I have been calling the primate sensory cortex) seems to create a model of this knowledge itself (areas of PSC get input from various areas of sensory neocortex). These new primate areas try to explain why the sensory neocortex believes food is over there, why an animal’s inner simulation of the external world is the way it is. An answer might be: Because last time I went over there I saw water, and hence when I simulate going back there, I see water in my imagination. Put slightly differently: Because I last saw water there, I now know that water is over there even though before I did not.
图 16.3
Figure 16.3
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
这些系统都是相互引导的。反射无需任何学习即可驱动效价反应，根据进化硬编码规则做出选择。脊椎动物的基底神经节和杏仁核可以根据这些反射在历史上得到强化的内容学习新的行为，根据最大化奖励做出选择。哺乳动物的 aPFC 可以学习这种行为的生成模型无模型行为并构建解释，根据想象的目标做出选择（例如，喝水）。这可以被认为是一阶模型。灵长类动物 gPFC 可以学习这种 aPFC 驱动行为的更抽象的生成模型（二阶模型），并构建意图本身的解释，根据心理状态和知识做出选择（我口渴；口渴时喝水感觉很好，当我模拟朝这个方向走时，我在模拟中发现了水，因此我想朝这个方向走）。
These systems are all bootstrapped on one another. Reflexes drive valence responses without any learning required, making choices based on evolutionarily hard-coded rules. The vertebrate basal ganglia and amygdala can then learn new behaviors based on what has historically been reinforced by these reflexes, making choices based on maximizing reward. The mammalian aPFC can then learn a generative model of this model-free behavior and construct explanations, making choices based on imagined goals (e.g., drinking water). This could be considered a first-order model. The primate gPFC can then learn a more abstract generative model (a second-order model) of this aPFC-driven behavior and construct explanations of intent itself, making choices based on mind states and knowledge (I’m thirsty; drinking water when thirsty feels good, and when I simulate going down this way, I find water in my simulation, hence I want to go in this direction).
哺乳动物的一级模型具有明显的进化优势：它使动物能够在行动之前间接地做出选择。但是，费尽心思开发二级模型有什么进化优势呢？为什么要对自己的意图和知识进行建模呢？
The mammalian first-order model has a clear evolutionary benefit: It enables the animal to vicariously play out choices before acting. But what is the evolutionary benefit of going through the trouble of developing a second-order model? Why model your own intent and knowledge?
塑造他人思维
Modeling Other Minds
以 Eric Brunet-Gouet 在 2000 年设计的连环画任务为例。人类参与者被展示几幅连环画，每幅包含三帧，并被要求猜测哪个第四帧的结局最有可能。连环画有两种类型——一种需要推断人物的意图才能正确猜出结局，另一种只需要理解物理因果关系。
Consider the comic-strip task devised by Eric Brunet-Gouet in 2000. Human participants were shown several comic strips, each containing three frames, and asked to guess which fourth-frame ending was most likely. There were two types of comic strips—one required inferring the intent of characters in order to correctly guess the ending, and the other required only understanding physical causal relationships.
Brunet-Gouet 在受试者完成这项连环画任务时对他们的大脑进行了 PET 扫描。他发现，大脑区域在每种类型的连环画中都会产生有趣的差异。当被问及需要理解人物意图的连环画时，这些大脑皮层中独有的灵长类区域（如 gPFC）会兴奋，而当被问及其他类型的连环画时则不会。激活后亮起。
Brunet-Gouet performed a PET scan of the subjects’ brains while they did this comic-strip task. He found an interesting difference in which brain regions got excited by each type of comic strip. When asked about comic strips that required understanding the intentions of the characters, but not when asked about other types of comic strips, these uniquely primate areas of neocortex, like gPFC, lit up with activation.
除了推断他人的意图之外，这些灵长类动物的大脑区域还会在需要推断他人知识的任务中被激活。一个著名的测试是莎莉-安测试：参与者会看到莎莉和安两个人之间发生的一系列事件。莎莉把一颗弹珠放进篮子里；莎莉离开；当莎莉不注意的时候，安把弹珠从篮子里移到附近的盒子里；莎莉回来了。然后问参与者：如果莎莉想玩她的弹珠，她会去哪里找？
In addition to inferring other people’s intent, these primate areas also get activated by tasks that require inferring other people’s knowledge. A famous test of this is the Sally-Ann test: Participants are shown a series of events occurring between two individuals, Sally and Ann. Sally puts a marble in a basket; Sally leaves; when Sally isn’t looking, Ann moves the marble from the basket to a nearby box; Sally comes back. The participant is then asked: If Sally wants to play with her marble, where will she look for it?
图 16.4 连环画任务示例：连环画 1 和 2 需要理解意图，连环画 3 不需要理解意图。连环画 1 的答案是 B（他想从窗户逃走）。连环画 2 的答案是 C（他想打开门）。连环画 3 的答案是 C（他会摔倒）。
Figure 16.4. Examples of comic-strip task: Comic strip 1 and 2 require understanding intent, comic strip 3 does not require understanding intent. Answer to comic strip 1 is B (he wants to escape through the window). Answer to comic strip 2 is C (he wants to open the door). Answer to comic strip 3 is C (he will fall down).
图片来自 Brunet 等人，2000 年；Völlm 等人，2006 年；以及与 Eric Brunet-Gouet 博士的私人通信。经 Brunet-Gouet 博士许可使用（私人通信）。
Images from Brunet et al., 2000; Völlm et al., 2006; and personal correspondence with Dr. Eric Brunet-Gouet. Used with permission of Dr. Brunet-Gouet (personal correspondence).
要正确回答这个问题，你需要意识到 Sally 的知识与你不同。你看到 Ann 把弹珠放进盒子里，而Sally没有。所以正确答案是 Sally 会看篮子，即使弹珠不在那里。Sally-Ann 测试有很多种形式，通常称为“错误信念测试”。人类通过这种错误信念测试的方式四岁时。当 fMRI 机器中的人类被给予错误信念测试中，灵长类动物的 gPFC 和 PSC 会亮起，人们的表现与激活程度。事实上，无数研究表明，这些独特的灵长类新皮层区域的许多区域被激活此类错误信念任务。
Answering this correctly requires realizing that Sally has different knowledge than you. While you saw Ann put the marble in the box, Sally did not. And so the right answer is that Sally will look in the basket, even though the marble is not there. There are many forms of the Sally-Ann test, generally referred to as “false-belief tests.” Humans pass such false-belief tests by the age of four. When humans in fMRI machines are given false-belief tests, the primate gPFC and PSC light up, and people’s performance correlates with the degree of activation. Indeed, across countless studies, it has been shown that many areas of these uniquely primate neocortical areas are specifically activated by such false-belief tasks.
Figure 16.5: The Sally-Ann test for theory of mind
照片来自 Frith，2003 年。已获得许可重复使用。
Photo from Frith, 2003. Reused with permission.
如果我们重新审视那些患有颗粒状前额叶损伤的神秘患者，并对他们进行心理理论测试，我们就会发现，从他们看似截然不同、微妙而又怪异的症状中，我们开始发现一个共同的主题。这类患者在解决像莎莉-安测试这样的错误信念测试方面表现更差；他们在识别其他人的情绪；他们很难感同身受与其他人的情绪作斗争区分谎言和笑话，难以识别失礼行为会冒犯某人努力地其他人的视角，以及欺骗别人的努力。
If we revisit our mysterious patients with granular prefrontal damage and test them for theory-of-mind tasks, we begin to see a common theme emerge from their seemingly disparate, subtle, and bizarre symptoms. Such patients are worse at solving false-belief tests like the Sally-Ann test; they are much worse at recognizing emotions in other people; they struggle to empathize with other people’s emotions, struggle to distinguish lies from jokes, struggle to identify a faux pas that would offend someone, struggle to take someone else’s visual perspective, and struggle to deceive others.
虽然上述所有研究都与人类大脑有关（尽管人类大脑的部分是从我们的灵长类祖先那里遗传下来的），但实验已经证实了同样的对非人类灵长类动物的影响。将一只猴子置于必须推理另一只猴子的意图或知识来解决任务的情境中，gPFC 就会亮起来，就像人类也是如此。如果猴子的 gPFC 受损，它们在执行此类任务时的表现就会受损，就像人类一样。
Although all of the above studies have to do with human brains (albeit the parts of human brains inherited from our primate ancestors), experiments have confirmed the same effects in nonhuman primates. Put a monkey in a situation where they have to reason about the intent or knowledge of another to solve a task, and gPFC lights up, just as it does in humans. Damage the gPFC in a monkey, and their performance on such tasks becomes impaired, just as it does in humans.
事实上，这些颗粒状前额叶区域的大小与理解他人的重要性相关，灵长类动物的社交网络规模。灵长类动物的粒状前额叶区域越大，其社会等级就越高。这种关系甚至在人类身上也能看到：人类的粒状前额叶区域越厚，其社交网络就越大，其在社交方面的表现就越好。心智理论任务。
And indeed, revealing their importance in understanding others, the size of these granular prefrontal areas is correlated with social-network size in primates. The bigger a primate’s granular prefrontal area, the higher in the social hierarchy it tends to be. This same relationship is even seen in humans: the thicker a human’s granular prefrontal areas, the larger his or her social network, and the better that person’s performance on theory-of-mind tasks.
这些新出现的灵长类动物大脑皮层区域似乎既是一个人对自己心智的模型，也是模仿其他思维。这两个看似不同的功能具有极为重叠甚至完全相同的神经基础，这一事实为这些新的灵长类动物结构的进化目的和机制提供了明显的线索。
These newly primate neocortical regions seem to be the locus of both one’s model of one’s own mind and the ability to model other minds. The fact that these two seemingly distinct functions have extremely overlapping, if not identical, neural substrates offers a glaring clue as to the evolutionary purpose and mechanism of these new primate structures.
用你的思维塑造他人的思维
Modeling Your Mind to Model Other Minds
早在柏拉图时代，就有一个关于人类如何理解他人思想的假设。该理论认为，我们首先了解自己的思想，然后利用对自己的理解来理解他人。这一古老思想的现代表述被称为“模拟理论”或“社会投射理论”。当我们试图理解为什么别人会做某事时，我们会想象自己处于他们的处境——拥有他们的知识和生活历史：“她对我大喊大叫可能是因为她明天要考试，她感到压力很大；我知道我压力大的时候会大喊大叫。”当我们试图理解别人会做什么时，我们会想象如果我们有他们的知识和背景，我们会处于他们的处境中会做什么：“我不认为詹姆斯会分享他的我相信詹姆斯看到乔治偷东西，而且我知道如果我看到我的朋友偷我的东西，我不会再和他分享了。”我们通过设身处地为他人着想来理解他人。
As far back as Plato, there has been a running hypothesis about how humans understand the minds of other humans. The theory is that we first understand our own minds and then use this understanding of ourselves to understand others. Modern formulations of this old idea are referred to as “simulation theory” or “social projection theory.” When we try to understand why someone else did something, we do so by imagining ourselves in their situation—with their knowledge and life history: “She probably yelled at me because she is stressed out about having a test tomorrow; I know I yell more when I am stressed.” When we try to understand what others will do, we imagine what we would do in their situation if we had their knowledge and their background: “I don’t think James will share his food with George anymore; I believe James saw George steal, and I know if I saw my friend steal from me, I wouldn’t share with him anymore.” We understand others by imagining ourselves in their shoes.
社会投射理论的最佳证据是，需要了解自己的任务和需要了解他人的任务都会激活并需要相同的灵长类动物独特的神经结构。在大脑中，推理自己的思想和推理他人的思想，同样的过程。
The best evidence for social projection theory is the fact that tasks that require understanding yourself and tasks that require understanding others both activate and require the same uniquely primate neural structures. Reasoning about your own mind and reasoning about other minds is, in the brain, the same process.
社会投射理论的证据也可以从儿童如何发展自我意识中找到。儿童时期自我意识的发展与儿童时期的发展密切相关心理理论。测试儿童自我意识的一种方法是镜子自我认知测试。在孩子的脸上涂上一抹东西，让他照镜子，看看他是否触摸了脸部的那个部分，意识到他正在看自己。孩子们往往要等到他们大约两岁。也是在这个时候，孩子们开始表现出对心理状态的原始理解，并开始使用诸如想要、希望和假装。大约三岁时，孩子们才意识到他们可以持有自己的错误信念；他们能够做出这样的陈述：“我以为那是一只鳄鱼。现在我知道这是鳄鱼。”直到这个年龄之后，大约四五岁的时候，孩子们才通过错误信念测试，比如莎莉安测试，尊重其他人。
Evidence for social projection theory can also be found in how children develop their sense of self. The childhood development of a sense of self is highly related to the childhood development of theory of mind. One way to test a child’s sense of self is the mirror self-recognition test. Put a smudge of something on a child’s face, let him look in the mirror, and see if he touches that part of his face, realizing that he is seeing himself. Children don’t tend to pass this test until they are about two years old. It is also around this time that children begin to exhibit a primitive understanding of mental states and begin to use words like want, wish, and pretend. It is later, around the age of three, when children realize they can hold their own false beliefs; they are able to make statements such as “I thought it was an alligator. Now I know it’s a crocodile.” It isn’t until after this point, around the age of four or five, when children pass false-belief tests, such as the Sally-Ann test, with respect to other people.
其他研究发现，儿童报告自己心理状态的能力与报告他人心理状态的能力之间存在很强的相关性——当儿童在某一方面表现更好时，他们往往会同时获得另一项技能更好。此外，一项技能的发育受损会削弱另一项技能；在社会隔离中长大的黑猩猩无法识别镜子中的自己。塑造自己的思想和塑造他人的思想是相互交织的。
Other studies have found a strong correlation between children’s ability to report on their own mental states and their ability to report on others’ mental states—when children get better at one, they tend to simultaneously get better at the other. Further, damage to the development of one impairs skills in the other; chimpanzees raised in social isolation are unable to recognize themselves in a mirror. Modeling your own mind and that of others is interwoven.
我们对自己的理解经常与对他人的理解交叉，这与我们为每个人重新设计一个共同系统的想法一致。例如，处于某种特定的情绪（如快乐或悲伤）会让你产生偏见，错误地推断这些状态在别人身上表现得很明显；口渴会让你产生偏见，错误地认为别人也比他们实际更渴。人们倾向于投射自己的性格特征影响他人。自我与他人的区别可能会变得混乱，如果我们通过将自己置于他人的处境中来理解他人，就会出现这种效果。
Our understanding of ourselves often gets cross-wired with our understanding of others, consistent with the idea that we are repurposing a common system for each. For example, being in a specific mood (such as happy or sad) makes you biased to incorrectly infer these states in others; being thirsty makes you biased to incorrectly think that others are thirstier than they are. People tend to project their own personality traits onto others. The self-other distinction can get cross-wired, an effect you would expect if we understand others by projecting ourselves into their situation.
对于我们最初的问题：心智理论是如何运作的？至少从概念上讲，一种可能性是，灵长类动物独有的新皮层区域首先构建了你自己内部模拟（换句话说，你的思想）的生成模型，然后使用这个模型尝试模拟其他人的想法。
To our original question: How might theory of mind work? One possibility, conceptually at least, might be that the uniquely primate neocortical areas first build a generative model of your own inner simulation (in other words, of your mind) and then use this model to try to simulate the minds of others.
我们处于抽象的境地——这很难成为构建具有心智理论的人工智能系统的详细算法蓝图。但是，通过首先建模自己的内部模拟、通过建模自己来建模他人来引导心智理论的想法提供了一个有趣的航点。今天，我们可以训练一个人工智能系统来观看人类行为的视频并预测人类下一步会做什么；我们向这些系统展示无数人类做事的视频，并告诉它们人类正在做什么的正确答案（“这是握手”；“这太跳跃了”。广告平台可以利用行为来预测人们下一步会买什么。我们确实有人工智能系统试图识别人们脸上的情绪（在系统接受了一堆照片的训练之后）我们无法用语言来描述人类的面部表情（比如，我们可以用表情来分类面孔）。但所有这些显然与人类（和其他灵长类动物）大脑的心理理论的复杂性相去甚远。如果我们想要人工智能系统和机器人能够与我们一起生活，了解我们是什么样的人，推断出我们不知道但想知道的东西，从我们说的话中推断出我们的意图，在我们告诉他们之前预测我们需要或想要什么，处理与人类群体的社会关系，了解他们所有的隐藏规则和礼仪——换句话说，如果我们想要真正的像人类一样的人工智能系统，心理理论无疑将成为该系统的重要组成部分。
We are in abstract land here—this is hardly a detailed algorithmic blueprint for how to build an AI system with theory of mind. But the idea of bootstrapping theory of mind by first modeling one’s own inner simulation, of modeling yourself to model others, provides an interesting waypoint. Today, we can train an AI system to watch videos of human behavior and predict what humans will do next; we show these systems endless videos of humans doing things and tell them the right answers of what the humans are doing (“This is handshaking”; “This is jumping”). Advertising platforms can use behavior to predict what people will buy next. And we do have AI systems that attempt to identify emotions in people’s faces (after the systems have been trained on a bunch of pictures of faces classified by emotions). But all this is clearly a far cry from the complexity of theory of mind in human (and other primate) brains. If we want AI systems and robots that can live alongside us, understand the type of people we are, deduce what we don’t know that we want to know, infer what we intend by what we say, anticipate what we need or want before we tell them, navigate social relationships with groups of humans, with all of their hidden rules and etiquettes—in other words, if we want true humanlike AI systems, theory of mind will undeniably be an essential component of that system.
事实上，心理理论可能是利用超级智能 AI 系统构建成功未来的最重要方面。如果超级智能 AI 系统无法推断我们所说的话的实际含义，我们就有可能进入一个反乌托邦的未来，届时 AI 系统会误解我们的要求，并可能带来灾难性的后果。我们将在突破 #5中进一步了解心理理论在向 AI 系统提出要求时的重要性。
In fact, theory of mind might be the most essential aspect of building a successful future with superintelligent AI systems. If superintelligent AI systems cannot infer what we actually mean by what we say, we risk entering a dystopian future where AI systems misinterpret our requests with possibly catastrophic consequences. We will learn more about the importance of theory of mind when making requests of AI systems in breakthrough #5.
在一群祖先灵长类动物的社会等级顶端是最上层的人有更多机会获得食物和配偶，最底层的人只能最后挑选食物，没有机会获得配偶。心智理论使每个灵长类动物都能爬上这个社会阶梯；它使他们能够管理自己的声誉并隐藏自己的过失；它使他们能够结成同盟，与冉冉升起的新星亲近，并加入强大家族的圈子；它使他们能够建立联盟和发动叛乱；它使他们能够缓和酝酿中的争端并在争吵后修复关系。与之前突破中出现的智力不同，心智理论并非源于在饥饿的捕食者或难以接近的猎物的危险中生存下来的需要，而是源于政治中更微妙、更尖锐的危险。
At the top of the social hierarchy of a troop of ancestral primates was more access to food and mates, and at the bottom was last pick of food and no access to mates. Theory of mind enabled each primate to climb this social ladder; it enabled them to manage their reputation and hide their transgressions; it enabled them to forge allyships, cozy up to rising stars, and kiss the ring of powerful families; it enabled them to build coalitions and stage rebellions; it enabled them to ameliorate brewing disputes and repair relationships after a tiff. Unlike the intellectual abilities that had emerged in the breakthroughs prior, theory of mind was not born from the need to survive the dangers of hungry predators or inaccessible prey, but instead from the subtler and far more cutting dangers of politics.
政治是《突破》第 4 期的起源故事，但它远非全部。正如我们将在接下来的两章中看到的那样，早期灵长类动物的心理理论被重新用于另外两种新能力。
Politics was the origin story of Breakthrough #4, but it is far from the entire story. As we will see in the next two chapters, theory of mind in early primates was repurposed for two other new abilities.
17
17
猴子锤和自动驾驶汽车
Monkey Hammers and Self-Driving Cars
简·古道尔简直不敢相信自己的眼睛。
JANE GOODALL COULD not believe her eyes.
那是 1960 年 11 月。几个月来，她一直在坦桑尼亚贡贝跟踪当地的一个黑猩猩部落。这些黑猩猩直到最近才开始接受她的存在，允许她靠近它们观察它们的自然栖息地。在那之前的几年里，古道尔与一位名叫路易斯·利基的肯尼亚古生物学家成为朋友；他最终提出派她去研究黑猩猩在自然栖息地的社交生活。但古道尔的第一个发现并不是关于它们的社交生活。
It was November 1960. For months, she had been following a local tribe of chimpanzees in Gombe, Tanzania. The chimps had only recently begun accepting her presence, allowing her to stay close enough to observe them in their natural habitat. In the years before, Goodall had befriended a Kenyan paleontologist by the name of Louis Leakey; he eventually offered to send her to study the social lives of chimpanzees in their natural habitat. But the first discovery Goodall would make was not about their social lives.
古道尔静静地坐在远处，她注意到两只黑猩猩，她给它们取名为“灰胡子大卫”和“歌利亚”，它们抓起细树枝，剥去树叶，然后把它们插进白蚁丘。当它们把树枝拔出来时，它们身上沾满了美味的白蚁，它们狼吞虎咽地吃着。它们在钓鱼。它们在使用工具。
As Goodall sat quietly some distance away, she noticed two chimpanzees that she had named David Greybeard and Goliath grabbing thin branches, stripping the leaves off them, and sticking them into a termite mound. When they pulled them out, they were covered in tasty termites, which they gobbled up. They were fishing. They were using tools.
长期以来，人们一直认为使用工具是人类独有的，但现在发现许多灵长类动物都会使用工具。猴子和猿类不仅使用棍子捕食白蚁，它们还会使用石头用来打开坚果、用草做牙线、用苔藓做海绵、用棍棒打碎蜂巢，甚至用树枝清洁耳朵。
It was long assumed that tool use was uniquely human, but tool use has now been found across many primates. Monkeys and apes not only use sticks to fish termites; they also use rocks to break open nuts, grass to floss, moss for sponges, clubs to smash beehives, and even twigs to clean their ears.
自从古道尔研究这些黑猩猩以来，动物界普遍存在使用工具的现象。大象用鼻子捡起树枝来拍打苍蝇和抓挠自己。猫鼬用铁砧砸开坚果。乌鸦用棍子刺幼虫。章鱼收集大贝壳制作盾牌。人们发现，濑鱼会用石头砸开蛤蜊，以获取到内部食物。
In the years since Goodall studied these chimps, tool use has been found all over the animal kingdom. Elephants pick up branches with their trunks to swat flies and scratch themselves. Mongooses use anvils to break open nuts. Crows use sticks to spear larvae. Octopuses gather large shells to make shields. Wrasses fish have been found to use rocks to break open clams to get to the inner food.
但灵长类动物使用工具的能力比其他动物更复杂。濑鱼、猫鼬和海獭已被证明会使用工具，但它们通常只有一种绝招。相比之下，黑猩猩群体通常表现出超过二十种不同的工具使用行为。此外，除了鸟类和大象之外，只有灵长类动物被证明会主动制造工具。黑猩猩会缩短、削尖并去除树枝上的叶子，然后再用它来捕捉白蚁。
But primate tool use is more sophisticated than tool use in other animals. Wrasses fish, mongooses, and sea otters have been shown to use tools, but they typically have only a single trick up their sleeve. In contrast, groups of chimpanzees often exhibit over twenty different tool-using behaviors. Further, with the possible exception of birds and elephants, only primates have been shown to actively manufacture their tools. A chimp will shorten, sharpen, and remove leaves from a stick before using it to fish for termites.
灵长类动物的工具使用也显示出不同社会群体之间的显著差异。不同种类的濑鱼虽然彼此之间没有接触，但使用石头的方式却相同。但灵长类动物的情况并非如此；同一灵长类动物的不同群体表现出令人惊讶的独特工具使用行为。古阿卢戈的黑猩猩制造白蚁钓鱼竿的方式不同比贡贝的要多。有些黑猩猩群会定期用石头敲开坚果，而有些则不会。有些黑猩猩群会用棍棒敲打蜂巢，而有些则不会。有些黑猩猩群会用多叶的树枝赶走苍蝇，而有些则不会。
Tool use in primates also shows a remarkable level of diversity across different social groups. Different species of wrasses fish use rocks in the same way despite having no contact with each other. But this is not the case with primates; different groups of the same species of primate exhibit surprisingly unique tool-using behaviors. The chimpanzees of Goualougo manufacture termite fishing rods differently than those in Gombe. Some groups of chimps regularly use rocks to open nuts; other groups don’t. Some groups of chimps use clubs to pound beehives; other groups don’t. Some groups use leafy twigs to swat away flies; others don’t.
如果早期灵长类动物大脑进化的驱动力是一场政治军备竞赛，那么为什么灵长类动物会成为独一无二的工具使用者呢？如果灵长类动物的新大脑区域是“设计”来支持心智理论的，那么灵长类动物独特的工具使用技能从何而来？
If the driver of brain evolution in early primates was a politicking arms race, why would primates be uniquely good tool users? If the new brain regions of primates were “designed” to enable theory of mind, then from where do the unique tool-using skills of primates emerge?
猴子镜子
Monkey Mirrors
1990 年，朱塞佩·迪·佩莱格里诺 (Giuseppe di Pellegrino)、莱昂纳多·福加西 (Leonardo Fogassi)、比托里奥·加莱塞 (Bittorio Gallese) 和卢西亚诺·法迪加 (Luciano Fadiga) 正在实验室里吃午餐。他们是帕尔马大学贾科莫·里佐拉蒂 (Giacomo Rizzolatti) 神经生理学实验室的成员；他们的任务是研究灵长类动物精细运动技能的神经机制。离午餐桌几英尺远的地方坐着一只猕猴，它是他们的研究对象。他们在它的大脑中放置了电极，以寻找哪些区域对特定类型的手部动作的执行有反应。他们发现，当猴子进行特定类型的手部动作时，运动前皮层的特定区域会被激活，有些抓握区域，握持区域，还有一些人因为流泪而受伤。但幸运的是，他们即将发现一些更了不起的东西。当实验室的一名成员拿起三明治咬了一口时，附近的扬声器发出了一声响亮的噼啪声。这声音不是来自连接火警或唱片机的扬声器，而是来自连接猴子大脑的扬声器。
In 1990, Giuseppe di Pellegrino, Leonardo Fogassi, Bittorio Gallese, and Luciano Fadiga were chowing down lunch in the lab. They were members of Giacomo Rizzolatti’s neurophysiology lab at the University of Parma; their mandate was to study the neural mechanisms of primate fine motor skills. A few feet from the lunch table sat a macaque monkey who was the object of their study. They had placed electrodes throughout its brain to search for which areas responded to the execution of specific types of hand movements. They had found specific areas of premotor cortex that activated when monkeys performed specific types of hand movements, some areas for grasping, others for holding, and others for tearing. But through lucky happenstance, they were about to discover something much more remarkable. As one of the lab members picked up a sandwich and bit into it, a loud crackle buzzed from a nearby speaker. The noise didn’t come from the speaker connected to the fire alarm or the record player, it came from the speaker connected to the monkey’s brain.
他们现在回忆起当时他们立即意识到有一件重要的事情刚刚发生的事情。他们只将电极连接到大脑皮层的运动区域，这些区域应该只在猴子自己进行特定的手部动作时才会激活。但在午餐时，尽管猴子一动不动，但当实验室的一名人类成员抓住自己的食物时，同一手部抓握区域却亮了起来。
As they now recall, they had an immediate perception that something important had just happened. They had attached electrodes only to the motor areas of neocortex, the areas that were supposed to activate only when the monkey itself performed specific hand movements. But at that moment during lunch, despite the monkey not moving at all, the same area for hand grasping lit up at the exact moment when one of the human lab members grasped their own food.
在尝试复制这种幻影三明治观看激活后，Rizzolati 的团队很快意识到，他们实际上发现了一些更普遍的东西：当他们的猴子观察人类执行运动技能时——无论是用两根手指捡起花生，用整只手抓住苹果，还是用嘴巴抓零食——猴子执行相同技能的运动神经元通常会被激活。换句话说，猴子大脑皮层运动前区和运动区中的神经元——控制猴子自身运动的神经元——不仅在执行特定的精细运动技能时被激活，而且在它们仅仅看着别人执行这些技能时也会被激活。Rizzolatti 将这些神经元称为“镜像神经元”。
After trying to replicate this phantom sandwich-watching activation, Rizzolati’s team quickly realized that they had, in fact, discovered something more general: when their monkey observed a human perform a motor skill—whether picking up a peanut with two fingers, grasping an apple with their full hand, or grasping a snack with their mouth—the monkey’s own motor neurons for performing that same skill would often activate. In other words, the neurons in the premotor and motor areas of a monkey’s neocortex—those that control a monkey’s own movements—not only activated when they performed those specific fine motor skills, but also when they merely watched others perform them. Rizzolatti called these “mirror neurons.”
在接下来的二十年里，里佐拉蒂的镜像神经元被发现存在于多种行为中（抓握、放置、握持、手指运动、咀嚼、咂嘴、伸出舌头），跨越大脑的多个区域（运动前皮层，顶叶、运动皮层）以及许多灵长类动物都有这种感觉。当一只灵长类动物观察另一只灵长类动物做某件事时，它的运动前皮层通常会模仿它所观察到的动作。
Over the subsequent twenty years, Rizzolatti’s mirror neurons have been found in numerous behaviors (grasping, placing, holding, finger movements, chewing, lip smacking, sticking one’s tongue out), across multiple areas of the brain (premotor cortex, parietal lobe, motor cortex), and across numerous species of primates. When a primate watches another primate do an action, its premotor cortex often mirrors the actions it is observing.
有多个竞争对手镜像神经元的解释。有人认为镜像神经元只不过是联想——运动神经元在响应与运动相关的任何提示时被激活。猴子看到自己的手臂在做抓握动作时它们选择做出抓握动作，因此当它们看到别人的手臂抓握东西时，其中一些相同的神经元当然会感到兴奋。其他人则认为镜像神经元代表着更基本的东西——也许镜像神经元是灵长类动物参与的机制心理理论。假设灵长类动物拥有某种聪明的机制，能够自动在心理上模仿它们在其他人身上看到的动作，然后通过模仿自己做这种行为，它们可以问“我为什么要这样做？”并试图推断另一只猴子或人类的意图。
There are multiple competing interpretations of mirror neurons. Some argue mirror neurons are nothing more than associations—motor neurons get activated in response to any cue that has been associated with a movement. Monkeys see their own arms make grasping movements when they choose to make grasping movements, so of course when they see someone else’s arms grasp things, some of the same neurons get a small kick of excitement. Others argue that mirror neurons represent something more fundamental—perhaps mirror neurons are the mechanism by which primates engage in theory of mind. The hypothesis is that primates have some clever mechanism for automatically mentally mirroring movements they see in others, and then by modeling themselves doing that behavior, they can ask “Why would I do this?” and try to deduce another monkey’s or human’s intentions.
其他人则有更中庸的解释。也许镜像神经元并没有某种自动的镜像机制；它们只是猴子碰巧想象自己在做它们想做的事情的线索。看看别人怎么做。镜像神经元并没有什么特别之处；它们只是猴子在看到你抓食物时会想到自己抓食物的证据。正如我们在第 12 章中看到的，在实际执行特定动作时激活的运动皮层区域也会在人们想象自己做同样的动作时被激活。
Others have a more middle-of-the-road interpretation. Perhaps mirror neurons don’t have some automatic mechanisms of mirroring; they are just clues that monkeys happen to be imagining themselves doing what they see someone else do. Mirror neurons aren’t anything special; they are just evidence that monkeys are thinking about themselves grasping food when they see you grasp food. And as we already saw in chapter 12, areas of the motor cortex that activate when actually performing specific movements also get activated when people imagine themselves doing those same movements.
这里有证据表明镜像神经元只是想象中的运动。猴子不需要直接观察运动来激活镜像神经元；只要给它们足够的信息，它们就能推断出正在执行什么动作。在猴子做出某种行为（例如拿起一颗花生并打算将其掰开）之前亮起的运动神经元，如果猴子只是听到花生掰开的声音，也会被激活打开（什么都没看到）。类似地，猴子捡起一个盒子时激活的神经元也会在猴子看到人类捡起一个被墙遮住的盒子时激活（但如果猴子知道没有墙，神经元就不会激活）。镜子神经元（例如，镜子神经元是猴子在墙后面的盒子里看到的镜子）如果镜子神经元只是自动镜子，那么在上述猴子没有直接观察行为的情况下，它们就不会被激活。然而，如果镜子神经元是想象行为的结果，那么每当某事触发猴子想象自己在做某事时，你就会看到这些镜子神经元被激活。
Here is some evidence that mirror neurons are just imagined movements. Monkeys don’t need to directly observe the movements for their mirror neurons to activate; they can merely be given sufficient information for them to infer what movements are being performed. The motor neurons that light up right before a monkey does a behavior (such as picking up a peanut with the intention of breaking it open) also activate if a monkey simply hears the peanut break open (without seeing anything). Similarly, the neurons in a monkey that activate when it picks up a box will also activate when the monkey sees a human presumably pick up a box that is obscured behind a wall (but the neurons don’t activate if monkeys know there is no box behind the wall). If mirror neurons were simply automatic mirrors, then they wouldn’t activate in the above cases where monkeys were not directly observing behaviors. However, if mirror neurons are the consequence of imagined behaviors, then whenever something triggers a monkey to imagine itself doing something, you would see these mirror neurons activate.
如果我们接受镜像神经元的解释，认为它是想象动作，那么这引出了一个问题，为什么猴子倾向于想象自己在做它们看到别人在做的事情呢？在心理上模拟别人的动作有什么意义呢？在第 12 章中，我们回顾了许多哺乳动物使用的运动模拟的一个好处：提前规划动作。这使猫能够快速计划将爪子放在哪里以走过平台，或者松鼠能够计划如何在不同的树枝之间跳跃。我们假设这可能是为什么哺乳动物具有如此出色的精细运动技能，而大多数爬行动物却非常笨拙的原因。但这个技巧与模拟别人的动作无关。
If we accept the interpretation of mirror neurons as the imagination of movements, then this begs the question why monkeys tend to imagine themselves doing what they see others doing. What is the point of mentally simulating movements you see in others? In chapter 12, we reviewed one benefit of motor simulation used by many mammals: planning movements ahead of time. This enables a cat to quickly plan where to place its paws to walk across a platform or a squirrel to plan how to jump between different branches. We hypothesized this might be why mammals have such great fine motor skills while most reptiles are woefully clumsy. But this trick has nothing to do with simulating movements you see in others.
模拟他人动作很有用的一个原因是，这样做有助于我们理解他们的意图。通过想象自己在做别人正在做的事情，你就可以开始理解他们为什么这样做：你可以想象自己在系鞋带或扣衬衫扣子，然后问自己“我为什么要做这样的事情？”，从而开始理解别人动作背后的潜在意图。最好的证据是一个奇怪的事实：在执行特定动作时有障碍的人，在理解别人执行相同动作的意图时也会有障碍。控制一组给定运动技能所需的运动前皮层亚区域与理解其他人执行相同运动技能的意图所需的亚区域相同。
One reason it is useful to simulate other people’s movements is that doing this helps us understand their intentions. By imagining yourself doing what others are doing, you can begin to understand why they are doing what they are doing: you can imagine yourself tying strings on a shoe or buttoning a shirt and then ask yourself “why would I do something like this?” and thereby begin to understand the underlying intentions behind other people’s movements. The best evidence for this is found in the bizarre fact that people with impairments in performing specific movements, also show impairments in understanding the intentions of those very same movements in others. The subregions of premotor cortex required for controlling a given set of motor skills are the same subregions required for understanding the intentions of others performing those same motor skills.
例如，在大脑皮层运动区域受损的患者中，动作产生（使用牙刷、梳子、叉子或橡皮擦等工具正确模仿的能力）和动作识别（正确选择与动作相匹配的模仿动作视频的能力）的障碍之间存在显著的相关性短语，如梳头）。那些自己刷牙很费劲的人往往不能很好地识别别人刷牙的情况。
For example, in patients with brain damage to motor areas of neocortex, there is a significant correlation between impairments to action production (the ability to correctly mime using tools such as toothbrushes, combs, forks, or erasers) and action recognition (the ability to correctly select a video of a mimed action that matches an action phrase, like combing hair). Individuals who struggle to brush their own teeth tend to be bad at recognizing teeth-brushing in others.
此外，暂时抑制人类的运动前皮层会削弱他们在观看某人捡起盒子的视频时正确推断盒子重量的能力（如果手臂能轻松拿起盒子，则表明盒子很轻，如果手臂一开始很吃力，必须调整位置才能获得更多杠杆作用，则表明盒子很重），但这不会影响他们通过观看视频推断球的重量的能力弹起。这表明，当人们看到其他人拿起一个盒子（“只有当盒子很重时，我才会把手臂转向那边”）。
Further, temporarily inhibiting a human’s premotor cortex impairs their ability to correctly infer the weight of a box when watching a video of someone picking it up (arms that easily pick it up suggest it is light, but arms that struggle at first and have to adjust their position to get more leverage suggest it is heavy), but it has no impact on their ability to infer a ball’s weight by watching a video of it bouncing on its own. This suggests that people mentally simulate themselves picking up a box when seeing someone else pick up a box (“I would turn my arm that way only if the box was heavy”).
这种对他人行为理解能力的下降并不是运动前皮层干扰的普遍影响，而是大脑无法模拟的身体部位所特有的。例如，暂时抑制运动前皮层的手部区域（模拟你自己的手部动作），不仅会削弱你自己手部动作的能力，还会削弱你识别模仿手部动作的能力（例如正确识别抓锤子或倒茶的模仿动作），但不会影响你识别模仿嘴部动作的能力（例如识别舔冰淇淋、例如，吃汉堡、吹蜡烛）。相反，暂时抑制运动前皮层的嘴部区域会损害你识别模仿嘴部动作的能力，但不会影响你识别手部动作的能力。
This impairment in understanding the actions of others is not some generalized effect of premotor cortex interference but is highly specific to the body parts that a brain is prevented from simulating. For example, temporarily inhibiting the hand area of your premotor cortex (which simulates your own hand movements), not only impairs your ability to perform your own hand movements, but also impairs your ability to recognize mimed hand movements (such as correctly recognizing mimed movements of grasping a hammer or pouring tea), but it has no impact on your ability to recognize mimed mouth movements (such as identifiying licking ice cream, eating a burger, blowing out a candle). Conversely, temporarily inhibiting the mouth area of your premotor cortex does impair your ability to recognized mimed mouth movements, while having no impact on your ability to recognize hand movements.
这表明，运动前皮层和运动皮层（模拟自己动作所需的大脑区域）也需要模拟他人的动作以理解他们的行为。但是，我们所说的理解，并不是指理解他人的情绪（饥饿、恐惧）或他人的知识（“比尔知道简藏了食物吗？”）。这些研究表明，运动前皮层专门参与理解他人行为的感觉运动方面——推断拿起一个盒子所需的力量或某人的工具类型旨在持有。
This suggests that the premotor cortex and motor cortex, the brain regions required to simulate your own movements, are also required to simulate the movements of others to understand their actions. But by understand, we don’t mean understanding others’ emotions (hungry, fearful) or others’ knowledge (“Does Bill know that Jane hid the food?”). These studies demonstrate that the premotor cortex is involved in specifically understanding the sensorimotor aspects of others’ behavior—inferring the strength required to pick up a box or the type of tool someone intended to hold.
但是，为什么正确识别你在他人身上观察到的行为的感觉运动方面很重要呢？意识到某人试图握住的工具或盒子的重量有什么好处呢？主要的好处是，它可以帮助我们通过观察学习新技能，就像它帮助早期灵长类动物一样。我们已经在第 14 章中看到，在脑海中演练动作可以提高实际执行动作时的表现。如果是这样的话，那么灵长类动物利用对他人观察来演练动作是有道理的。
But why does it matter to correctly identify the sensorimotor aspects of the behaviors you observe in others? What benefit does it provide to realize the tool someone is trying to hold or the weight of a box? The main benefit is that it helps us, as it helped early primates, learn new skills through observation. We already saw in chapter 14 that mentally rehearsing actions improves performance when actually performing actions. If this is the case, then it would make sense for primates to use their observations of others to rehearse actions.
假设你让一名新手吉他手使用 fMRI 机器，并要求他们通过观看专业吉他手弹奏该和弦的视频来学习吉他和弦。假设你比较他们在以下情况下的大脑活动两种情况，第一种是当他们观察一个他们还不知道的和弦时，第二种是当他们观察一个他们已经知道如何演奏的和弦时。结果：当他们观察一个他们还不知道的和弦时，他们的运动前皮层变得比观察他们已经知道如何演奏的和弦时更加活跃。
Suppose you put a novice guitar player in an fMRI machine and ask them to learn a guitar chord by watching a video of an expert guitarist playing that chord. And suppose you compare their brain activation under two conditions, the first being when they observe a chord they don’t know yet and the second when they observe a chord they already know how to play. The result: when they observe a chord they do not yet know, their premotor cortex becomes way more activated than when they observe a chord they already know how to play.
但是，运动前皮层在通过观察学习新技能时会特别活跃，这一事实并不能证明学习新技能必须通过观察。假设你让一个人观看两个不同的视频。在第一个视频中，他看到一只手在按键盘上的特定按钮，他被要求模仿这些手势并按自己版本的相同键盘。在另一个视频中，他看到一个红点移动到键盘上的不同按钮，他被要求按自己键盘上的相同按钮。如果你在这个任务期间暂时抑制他的运动前皮层，他在模仿手势方面会受到特别的损害，但在模仿手势方面表现正常跟随红点。运动前激活不仅与模仿学习相关；至少在某些情况下，它似乎是模仿学习所必需的。在这里我们可以开始解开为什么灵长类动物是如此优秀的工具使用者。
But the fact that premotor cortex becomes uniquely activated when trying to learn a new skill through observation, does not prove that it is required to learn a new skill through observation. Suppose you ask a human to watch two different videos. In the first video, he sees a hand pushing specific buttons on a keyboard, and he is asked to imitate those hand movements and push his own version of the same keyboard. In the other video, he sees a red dot move to different buttons on a keyboard, and he is asked to push the same buttons on his own keyboard. If you temporarily inhibit his premotor cortex during this task, he becomes specifically impaired at imitating hand motions but performs normally at following the red dots. Premotor activation is not just correlated with imitation learning; it seems to be, at least in some contexts, necessary for imitation learning. And here we can begin to unravel why primates are such great tool users.
传播性胜过独创性
Transmissibility Beats Ingenuity
想想所有与使用工具有关的巧妙运动技能：打字、开车、刷牙、打领带或骑自行车。这些技能中有多少是你自己掌握的？我敢打赌，几乎所有这些技能都是通过观察他人获得的，而不是靠你自己的独立智慧获得的。非人类灵长类动物使用工具的起源也是一样的。
Think about all the clever motor skills related to using tools: typing, driving, brushing your teeth, tying a tie, or riding a bicycle. How many of these skills did you figure out on your own? I’m going to bet that practically all these skills were acquired by observing others, not by your own independent ingenuity. Tool use in nonhuman primates originates the same way.
群体中的大多数黑猩猩使用相同的工具技巧，并不是因为它们都独立想出了相同的技巧，而是因为它们通过相互观察而学习。一只年轻的黑猩猩花在观察母亲使用白蚁捕捞工具或蚂蚁浸渍工具上的时间，是它学习每项技能的年龄的重要预测因素；它观察得越多，它就越早学会学会了如何做到这一点。如果没有其他人的传递，大多数黑猩猩永远无法自己弄清楚如何使用工具；事实上，如果一只小黑猩猩在五岁之前没有通过观察其他人学会如何砸开坚果，它就不会掌握以后生活中的技能。
Most chimps in a group use the same tool techniques not because they all independently came up with the same trick but because they learned by observing each other. The amount of time a young chimp spends watching its mother using termite fishing tools or ant dipping tools is a significant predictor of the age at which it will learn each skill; the more it watches, the earlier it learns to do it. Without transmission from others, most chimps never figure tool use out on their own; in fact, a young chimp that doesn’t learn, through observing others, to crack nuts by the age of five will not acquire the skill later in life.
实验室实验已经证实了非人类灵长类动物的技能传递。在 1987 年的一项研究中，一群年轻的黑猩猩被给予一个 T 形耙子，它可以插入笼子并用来抓取远处的食物。一半的黑猩猩观察了成年黑猩猩使用该工具，另一半则没有。观察了成年专家演示的那组年轻黑猩猩弄清楚了如何使用该工具，而没有看到专家演示的那组年轻黑猩猩从未弄清楚如何使用该工具（尽管他们非常积极，因为他们可以看到笼子里的食物）。
Skill transmission in nonhuman primates has been shown in lab experiments. In a 1987 study, a group of young chimps were given a T-bar rake that could be inserted through a cage and used to grasp faraway food. Half the chimps observed an adult chimpanzee use the tool, and the other half did not. The group of young chimps that watched the expert adult figured out how to use the tool, while the group that saw no expert demonstrations never figured out how to use the tool (despite being highly motivated because they could see the food in the cage).
这些技能可以传授给整个灵长类动物群体。请看以下研究。实验人员暂时从他们的群体中带走一只黑猩猩、卷尾猴或狨猴，并教它一项新技能。这些个体被教导技能，例如用棍子戳食物分配装置以正确的方式，以特定的方式滑动打开门获取食物的方法，拉抽屉里放食物，或者打开人造水果。在教授这项新技能后，实验人员将已经熟练掌握的灵长类动物重新引入他们的小组。在一个月内，几乎整个小组都在使用这些相同的技巧，而从未有人教授过这项技能的小组永远不会弄清楚如何以相同的方式使用这些工具。这些技能最初只教给一个人，后来被传授给传承了多代。
These skills can propagate throughout an entire group of primates. Consider the following studies. Experimenters temporarily took an individual chimpanzee, capuchin monkey, or marmoset from their group and taught it a new skill. These individuals were taught skills such as using a stick to poke a food-dispensing device in the right way, sliding open a door in a specific way to get food, pulling a drawer to get food, or opening an artificial fruit. After teaching this new skill, experimenters reintroduced the now skilled primate back to their group. Within a month, almost the entire group was using these same techniques, whereas groups that never had a member taught the skill never figured out how to use the tools in the same way. And such skills, originally taught to only a single individual, were passed down through multiple generations.
使用工具的能力与创造力无关，而与传承性有关。如果传承性经常发生，创造力必定只会发生一次；如果一个群体中至少有一个成员知道如何制造和使用捕白蚁棒，那么整个群体都可以掌握这项技能，并不断将其传承给后代。
The ability to use tools is less about ingenuity and more about transmissibility. Ingenuity must occur only once if transmissibility occurs frequently; if at least one member of a group figures out how to manufacture and use a termite-catching stick, the entire group can acquire this skill and continuously pass it down throughout generations.
但仅仅因为灵长类动物具有在它们之间传递运动行为的能力（通过观察学习），就断定它们是唯一擅长使用工具的动物，这是不准确的。许多动物使用工具的能力比灵长类动物差得多，甚至那些根本不使用工具的动物，也参与了观察学习。老鼠可以通过观察另一只老鼠推杠杆来学习推杠杆取水。杠杆并获取水。猫鼬将采用开口的海豚可以模仿父母的动作。其他海豚或人类。狗可以通过观察另一只狗，学习如何用爪子拉动杠杆来获取食物狗也能完成这一行为。甚至鱼和爬行动物也能观察到导航路径被同类其他成员所吸收，并学会吸收那些相同的导航路径。
But it would be inaccurate to conclude that primates are uniquely good tool users simply because of their ability to transmit motor behaviors among themselves—to learn by observation. Many animals who are much worse tool users than primates, even those who don’t use tools at all, also engage in observational learning. Rats can learn to push a lever to get water by watching another rat push a lever and get water. Mongooses will adopt the egg-opening technique of their parents. Dolphins can be trained to imitate movements they see in other dolphins or humans. Dogs can learn how to pull a lever with their paw to get food by watching another dog perform the act. Even fish and reptiles can observe the navigational paths taken by other members of their own species and learn to take those same navigational paths.
选择已知技能 SELECTING KNOWN SKILLS 通过观察 THROUGH OBSERVATION |
获得新技能 ACQUIRING NOVEL SKILLS 通过观察 THROUGH OBSERVATION |
许多哺乳动物 Many mammals 章鱼 Octopuses 鱼 Fish 爬行动物 Reptiles |
灵长类动物 Primates 一些鸟 Some birds |
但是灵长类动物的观察学习与大多数其他哺乳动物有所不同。如果一只猫鼬父母倾向于用嘴巴打开鸡蛋，那么它的后代也会这么做；如果一只猫鼬父母倾向于通过投掷打开鸡蛋，那么它的后代也会这么做。但这些幼年猫鼬并没有通过观察获得新技能；它们只是改变了它们倾向于使用的技巧——所有的幼年猫鼬都表现出咬和投掷打开鸡蛋的技巧。小猫只有在看到母亲在猫砂盆里撒尿时才会学会，但所有的小猫都知道如何撒尿。鱼不是通过观察来学习游泳的；它们只是通过观察改变路径。在所有这些情况下，动物都没有使用观察学习来获得新技能；它们只是根据看到另一种动物做同样的事情来选择一种已知行为。
But there is a difference between observational learning in primates relative to most other mammals. If a parent mongoose tends to break open eggs with its mouth, then so does its offspring; if a parent mongoose tends to break open eggs by throwing, then so does its offspring. But these children mongooses aren’t acquiring a novel skill by watching; they are merely changing which technique they tend to use—all children mongooses exhibit both biting and throwing tricks for opening eggs. Kittens learn to pee in litter boxes only if exposed to their mothers doing it, but all kittens know how to pee. Fish don’t learn how to swim by watching; they merely change their paths by watching. In all these cases, animals aren’t using observational learning to acquire novel skills; they are merely selecting a known behavior based on seeing another do the same thing.
通过观察选择已知行为可以通过简单的反射来实现：乌龟可能会反射性地朝其他乌龟注视的方向看；鱼可能会反射性地跟随其他鱼。当老鼠观察到另一只老鼠推杠杆时，它可以模拟自己推杠杆（它已经知道如何做），此时这只老鼠会意识到如果这样做，它会得到水。但是通过观察获得一种全新的运动技能可能需要全新的机器，或者至少会从中受益匪浅。
Selecting a known behavior through observation can be accomplished with simple reflexes: A tortoise may have a reflex to look in the direction that other tortoises are looking; a fish may have a reflex to follow other fish. A mouse can simulate itself pushing a lever (something it already knows how to do) when it observes another mouse pushing a lever, at which point this mouse will realize that it will get water if it does this. But acquiring an entirely novel motor skill by observation may have required, or at least hugely benefited from, entirely new machinery.
为什么灵长类动物使用锤子而老鼠却不用
Why Primates Use Hammers but Rats Don’t
通过观察获得新技能需要心理理论，而通过观察选择已知技能则不需要。有三个原因为什么会这样。心理理论对于通过观察获得新技能必不可少的第一个原因是，它可能使我们的祖先能够积极地教学。要让技能在人群中传播，你不需要老师——新手的尽职观察就可以了。但积极教学可以大大改善技能的传递。想想如果没有老师放慢速度并指导你完成每个步骤，你不得不自己解读步骤，看着人们快速系鞋带，而不顾你的学习，那么学习系鞋带会困难得多。
Acquiring novel skills through observation required theory of mind, while selecting known skills through observation did not. There are three reasons why this was the case. The first reason why theory of mind was necessary for acquiring novel skills by observation is that it may have enabled our ancestors to actively teach. For skills to be transmitted through a population, you don’t need teachers—dutiful observation by novices will do. But active teaching can substantially improve the transmission of skills. Think about how much harder it would have been to learn to tie your shoes if you didn’t have a teacher who slowed down and walked you through each step, and instead you had to decipher the steps on your own by watching people rapidly tie their shoes with no regard for your learning.
只有借助心智理论，教学才有可能。教学需要了解另一个心智不知道什么，以及哪些演示有助于以正确的方式操纵另一个心智的知识。虽然除了人类之外，是否有其他灵长类动物会教学仍存在争议，但近年来，越来越多的证据开始支持非人类灵长类动物确实会教学的观点。积极地互相教导。
Teaching is possible only with theory of mind. Teaching requires understanding what another mind does not know and what demonstrations would help manipulate another mind’s knowledge in the correct way. While it is still controversial whether any primates other than humans teach, in recent years evidence is beginning to accumulate in favor of the idea that nonhuman primates do, in fact, actively teach one another.
20 世纪 90 年代，灵长类动物学家 Christophe Boesch 报告称，她观察到黑猩猩母亲在幼崽周围慢动作敲坚果。这些母亲会定期查看，以确保孩子在集中注意力。Boesch 报告称，黑猩猩母亲会通过取出坚果、清洁砧座，然后放回坚果来纠正幼崽的错误。他还报告称，母亲会调整锤子的方向。他们的年轻人的手。
In the 1990s, the primatologist Christophe Boesch reportedly observed chimp mothers performing nut cracking in slow motion specifically around their young. These mothers periodically looked over to ensure their child was paying attention. Boesch reported chimp mothers correcting mistakes in youngsters by removing a nut, cleaning the anvil, then placing the nut back. He also reported mothers reorienting a hammer in the hands of their young.
研究发现，猴子在还没有学会这种技能的幼崽周围会夸张地“使用牙线”，就好像它们在放慢速度俯下身来帮忙教导。擅长白蚁捕鱼的黑猩猩通常会带两根棍子参加捕鱼活动，并直接将其中一根递给幼崽。如果幼崽没有带棍子，黑猩猩甚至会将自己的棍子折成两半，将一半递给幼崽。如果幼崽开始看起来像是在做某项任务时遇到困难，母亲就会积极地与他们交换工具。使用工具的过程越复杂，母亲越有可能给予对年轻人来说是一种工具。
Monkeys have been found to exaggerate their “flossing” specifically around youngsters that have not yet learned this skill, as if they are slowing down to help teach. Chimps that are skilled termite fishers will often bring two sticks to a fishing activity and directly hand one to a youngster. A chimp will even break its own stick in two and give the youngster half if the child shows up without a stick. If a child starts to seem like it is struggling with a task, the mother will actively swap tools with them. And the more complex the tool-using process, the more likely a mother is to give a tool to a youngster.
心理理论对于通过观察学习新运动技能必不可少的第二个原因是，它使学习者能够长时间专注于学习。一只老鼠可以看到另一只老鼠推杠杆，几分钟后它自己也推了杠杆。但一只黑猩猩孩子会看着它的母亲用铁砧敲开坚果，练习了好几年却一无所获，直到它开始掌握这项技能。黑猩猩幼崽不断尝试学习，但没有任何短期奖励。
The second reason why theory of mind was necessary for learning novel motor skills through observation is that it enabled learners to stay focused on learning over long periods. A rat can see another rat push a lever and a few moments later push the lever itself. But a chimpanzee child will watch its mother use anvils to break open nuts and practice this technique for years without any success before it begins to master the skill. Chimp children continually attempt to learn without any near-term reward.
幼年黑猩猩这样做可能只是因为它们觉得模仿本身就是一种奖励，但另一种可能性是，心智理论使新手能够识别复杂技能的意图，从而使它们有很强的动力继续尝试掌握它。心智理论使幼年黑猩猩意识到，当母亲用棍子得到食物时，它却用棍子得不到食物，是因为母亲拥有它尚未掌握的技能。这促使它们不断获得技能，即使需要很长时间才能掌握。另一方面，当老鼠模仿行为时，如果它的行为没有带来短期奖励，它就会很快放弃。
It is possible chimp children do this simply because they find imitation rewarding on its own, but another possibility is that theory of mind enables novices to identify the intent of a complex skill, which makes them highly motivated to keep trying to adopt it. Theory of mind enables a chimp child to realize that the reason it is not getting food with its stick while its mother is getting food is that its mother has a skill it does not yet have. This enables a continuous motivation to acquire the skill, even if it takes a long time to master. When a rat imitates behaviors, on the other hand, it will quickly give up if its actions don’t lead to a near term reward.
心理理论对于通过观察学习新运动技能必不可少的第三个也是最后一个原因是，它使新手能够区分专家的有意和无意的动作。如果一个人知道另一个人试图通过每个动作完成什么，观察学习会更有效。如果你看着你妈妈系自己的鞋带，而你不知道她的哪些动作是有意的，哪些是无意的，那么很难分辨出应该跟随哪些动作。如果你意识到她的意图是系好鞋带，当她滑倒时那是意外，而她的坐姿和头部的角度都是技能的无关方面，那么通过观察学习这项技能对你来说就会容易得多。
The third and final reason why theory of mind was necessary for learning novel motor skills through observation was that it enabled novices to differentiate between the intentional and unintentional movements of experts. Observational learning is more effective if one is aware of what another is trying to accomplish with each movement. If you watched your mother tie her own shoes and you had no idea what aspects of her movements were intentional versus accidental, it would be quite hard to decipher which movements to follow. If you realized that her intention was to get her shoes tied, that when she slipped it was an accident, and that both the way she is seated and the angle of her head are irrelevant aspects of the skill, it would be much easier for you to learn the skill through observation.
这确实是黑猩猩学习的方式。考虑以下实验。一只成年黑猩猩被允许观察人类实验者打开一个拼图盒来获取食物。在打开拼图盒所需的一系列动作中，实验者做了几个不相关的动作，比如敲击魔杖或旋转盒子。然后，黑猩猩有机会打开拼图盒自己获取食物。令人惊讶的是，黑猩猩并没有完全模仿实验者的每一个动作，而是只模仿打开拼图盒所需的动作，然后跳过了不相关的步骤。
This is indeed how chimpanzees learn. Consider the following experiment. An adult chimpanzee was allowed to observe a human experimenter open a puzzle box to get food. Throughout the sequence of actions needed to open the puzzle box, the experimenter did several irrelevant actions, such as tapping a wand or rotating the box. Then the chimp was given a chance to open the puzzle box to get food itself. Amazingly, the chimp didn’t exactly copy every movement of the experimenter, instead it copied only the necessary movements to open the puzzle box and skipped the irrelevant steps.
理解动作的意图对于观察学习的发挥作用至关重要；它使我们能够过滤掉多余的动作并提取技能的精髓。
Understanding the intentions of movements is essential for observational learning to work; it enables us to filter out extraneous movements and extract the essence of a skill.
机器人模仿
Robot Imitation
1990 年，卡内基梅隆大学的研究生 Dean Pomerleau 和他的导师 Chuck Thorpe 开发了一个自动驾驶汽车的人工智能系统。他们将其称为 ALVINN（神经网络中的自动陆地车辆）。ALVINN 接收来自汽车周围的视频片段，可以自行操控汽车保持在真实高速公路上的车道内。之前也曾有人尝试过这样的自动驾驶汽车，但它们的速度非常慢，经常每隔几秒钟就会停顿一次；Thorpe 团队的原始版本的自动驾驶汽车每小时只能行驶四分之一英里，因为它需要进行大量思考。ALVINN 的速度要快得多，事实上，ALVINN 的速度非常快，以至于 ALVINN 与其他司机一起在真实的高速公路上成功地将 Pomerleau 从匹兹堡开到了五大湖区。
In 1990, a graduate student at Carnegie Mellon named Dean Pomerleau and his adviser Chuck Thorpe built an AI system to autonomously drive a car. They called it ALVINN (Autonomous Land Vehicle in a Neural Network). ALVINN was fed video footage from around a car and could—on its own—steer the car to stay within a lane on a real highway. There had been previous attempts at autonomous cars like this, but they were very slow, often pausing every few seconds; the original version of the autonomous car in Thorpe’s group could go only a quarter of a mile per hour due to how much thinking it had to do. ALVINN was much faster, so fast, in fact, that ALVINN successfully drove Pomerleau from Pittsburgh to the Great Lakes on a real highway with other drivers.
为什么 ALVINN 成功了，而之前的尝试却失败了？与之前制造自动驾驶汽车的尝试不同，ALVINN 没有被训练识别物体、规划未来运动或了解其在空间中的位置。相反，ALVINN 通过做一些更简单的事情超越了其他人工智能系统：它通过模仿人类驾驶员来学习。
Why was ALVINN successful while previous attempts had failed? Unlike previous attempts to build a self-driving car, ALVINN was not taught to recognize objects or plan its future movements or understand its location in space. Instead, ALVINN outperformed these other AI systems by doing something simpler: It learned by imitating human drivers.
Pomerleau 训练 ALVINN 的方式如下：他在车上安装了一个摄像头，记录开车时拍摄的视频和方向盘的位置。*然后，Pomerleau 训练了一个神经网络，将道路图像映射到他选择的相应转向位置。换句话说，他训练 ALVINN直接模仿Pomerleau 的动作。令人惊讶的是，仅观察了几分钟，ALVINN 就能够很好地自行驾驶汽车。
Pomerleau trained ALVINN the following way: He put a camera on his car and recorded both the video and the position of his steering wheel while he drove around.* Pomerleau then trained a neural network to map the image of the road to the corresponding steering position that he had selected. In other words, he trained ALVINN to directly copy what Pomerleau would do. And remarkably, after only a few minutes of observation, ALVINN was quite good at steering the car on its own.
但随后，波默洛遇到了一个问题。他很快发现，这种模仿学习方法（直接模仿专家的行为）存在一个关键缺陷。只要 ALVINN 犯了哪怕是很小的错误，它都完全无法恢复。小错误会迅速导致驾驶的灾难性失败，通常会完全偏离道路。问题是 ALVINN 只接受过正确驾驶的训练。它从未见过人类从错误中恢复过来，因为它根本就没见过错误。事实证明，直接模仿专家的行为是一种危险而脆弱的模仿学习方法。
But then Pomerleau hit a snag; it quickly became clear that this approach to imitation learning—of copying expert behavior directly—had a critical flaw. Whenever ALVINN made even small errors, it was completely unable to recover. Small mistakes would rapidly cascade into catastrophic failures of driving, often veering entirely off the road. The problem was that ALVINN was trained only on correct driving. It had never seen a human recover from a mistake because it had never seen a mistake in the first place. Directly copying expert behaviors turned out to be a dangerously brittle approach to imitation learning.
在机器人技术中，有许多策略可以克服这个问题。但其中两种策略与灵长类动物模仿学习的方式有着明显的相似之处。第一种是模仿师生关系。除了训练人工智能系统直接模仿专家之外，如果专家也与人工智能系统一起驾驶并纠正其错误会怎样？2009 年，卡内基梅隆大学的 Stephane Ross 和他的顾问 Drew Bagnell 首次尝试了这种方法。他们教人工智能系统在模拟的马里奥赛车环境中驾驶。Ross 没有录制自己的驾驶过程，然后训练系统模仿，而是在马里奥赛车赛道上驾驶，与人工智能系统交换对汽车的控制权。起初，Ross 会进行大部分驾驶，然后控制权会暂时移交给人工智能系统，它犯的任何错误都会被 Ross 迅速纠正。随着时间的推移，Ross 将更多的控制权交给人工智能系统，直到它能够独立驾驶得很好。
There are numerous strategies to overcome this problem in robotics. But two in particular draw conspicuous parallels to how primates seem to make imitation learning work. The first is to emulate a teacher-student relationship. In addition to training an AI system to directly copy an expert, what if the expert also drove alongside the AI system and corrected its mistakes? One of the first attempts to do this was by Stephane Ross and his adviser Drew Bagnell at Carnegie Mellon in 2009. They taught an AI system to drive in a simulated Mario Kart environment. Instead of recording himself drive and then training a system to imitate it, Ross drove around the Mario Kart track and traded control over the car with the AI system. At first, Ross would do most of the driving, then control would be passed to the AI system for a moment, and any mistakes it made would be quickly recovered by Ross. Over time, Ross gave more control to the AI system until it was driving well on its own.
这种主动教学策略效果非常好。当只直接模仿驾驶（比如 ALVINN 的训练）时，Ross 的 AI 系统在经过一百万帧专家数据后仍然会撞车。相比之下，采用这种新的主动教学策略后，他的 AI 系统在仅仅训练了 100 万帧后就几乎完美地驾驶了。几圈。这就像一只黑猩猩老师纠正正在学习新技能的孩子的动作一样。一只黑猩猩妈妈看着她的女儿试图把一根棍子插进白蚁丘，当孩子挣扎的时候，妈妈会试图纠正她。
This strategy of active teaching worked fantasically. When only directly copying driving (like ALVINN was trained), Ross’s AI system was still crashing cars after a million frames of expert data. In contrast, with this new strategy of active teaching, his AI system was driving almost perfectly after only a handful of laps. This is not unlike a teacher chimpanzee correcting the movements of a child learning a new skill. A chimp mother watches her daughter try to insert a stick into a termite mound, and when the child is struggling, the mother attempts to correct her.
机器人模仿学习的第二种方法是这被称为“逆向强化学习”。如果人工智能系统不直接复制人类根据道路图片做出的驾驶决策，而是首先尝试识别人类驾驶决策的意图，结果会怎样？
The second approach to imitation learning in robotics is called “inverse reinforcement learning.” Instead of trying to directly copy the driving decisions that a human makes in response to a picture of the road, what if an AI system first attempts to identify the intent of the human’s driving decisions?
2010 年，Pieter Abbeel、Adam Coates 和 Andrew Ng 利用逆向强化学习让 AI 系统自主飞行，展示了其强大功能遥控直升机。驾驶直升机，即使是遥控直升机，也很难；直升机不稳定（小错误可能会迅速导致坠机），需要不断调整才能保持在空中，并且需要同时正确平衡多个复杂输入同时（顶部叶片的角度、尾桨叶片的角度、直升机机身的倾斜方向等等）。
In 2010, Pieter Abbeel, Adam Coates, and Andrew Ng demonstrated the power of inverse reinforcement learning by using it to get an AI system to autonomously fly a remote-controlled helicopter. Flying a helicopter, even a remote-controlled one, is hard; helicopters are unstable (small mistakes can rapidly cascade into a crash), require constant adjustments to keep them in the air, and require correctly balancing multiple complex inputs at the same time (angle of top blades, angle of tail rotor blades, tilt orientation of the helicopter body, and more).
吴恩达和他的团队并不想让人工智能系统简单地驾驶直升机，他们希望它能够执行只有最优秀的人类专家才能完成的杂技任务：原地翻转而不跌倒、向前滚动、倒飞、空中回环等等。
Ng and his team didn’t want to get an AI system to simply fly a helicopter, they wanted it to perform acrobatic tasks, those that only the best human experts could perform: flipping in place without falling, rolling while moving forward, flying upside down, performing arial loops, and more.
他们的方法之一是标准的模仿学习。吴恩达和他的团队在人类专家表演这些杂技时，记录了他们向遥控器输入的信息。但他们没有训练人工智能系统直接模仿人类专家（这种方法行不通），而是训练人工智能系统首先推断专家的预期轨迹，推断出人类似乎想要做什么，然后人工智能系统学会追寻这些预期轨迹。这种技术被称为“逆向强化学习”，因为这些系统首先尝试学习他们认为熟练专家正在优化的奖励函数（即他们的“意图”），然后这些系统通过反复试验进行学习，使用这个推断出的奖励函数奖励和惩罚自己。逆向强化学习算法从观察到的行为开始，并产生自己的奖励函数；而在标准强化学习中，奖励函数是硬编码的，不是学习的。即使是专业飞行员驾驶这些直升机，他们也会不断从小错误中恢复过来。通过首先尝试识别预期的轨迹和动作，吴恩达的人工智能系统既可以过滤掉专家的无关错误，也可以纠正自己的错误。利用逆向强化学习，到 2010 年，他们成功训练了一个人工智能系统，让其能够自主地驾驶直升机进行空中特技飞行。
Part of their approach was standard imitation learning. Ng and his team recorded human expert inputs to the remote control as they performed these acrobatic tricks. But instead of training the AI system to directly copy the human experts (which didn’t work), they trained the AI system to first infer the intended trajectories of the experts, inferring what it seemed like the humans were trying to do, and then the AI system learned to pursue those intended trajectories. This technique is called “inverse reinforcement learning” because these systems first try to learn the reward function they believe the skilled expert is optimizing for (i.e., their “intent”), and then these systems learn by trial and error, rewarding and punishing themselves using this inferred reward function. An inverse reinforcement learning algorithm starts from an observed behavior and produces its own reward function; whereas in standard reinforcement learning the reward function is hard-coded and not learned. Even when expert pilots flew these helicopters, they were continually recovering from small mistakes. By first trying to identify the intended trajectories and movements, Ng’s AI system was both filtering out extraneous mistakes of experts, and correcting its own mistakes. Using inverse reinforcement learning, by 2010 they successfully trained an AI system to autonomously perform arial aerobatics with a helicopter.
在机器人模仿学习。但事实上，逆向强化学习（即人工智能系统推断观察到的行为的意图）似乎是观察学习发挥作用的必要条件，至少在某些任务中是如此，这一事实支持了这样一种观点，即心智理论（即灵长类动物推断观察到的行为的意图）是观察学习和工具技能在它们之间传递所必需的。机器人专家的聪明才智和进化的迭代都集中在类似的解决方案上，这不太可能是巧合；新手不能仅仅通过新手在观察高手的动作的同时，也必须洞察高手的内心。
There is still much work to be done when it comes to imitation learning in robotics. But the fact that inverse reinforcement learning (whereby AI systems infer the intent of observed behavior) seems necessary for observational learning to work, at least in some tasks, supports the idea that theory of mind (whereby primates infer the intent of observed behavior) was required for observational learning and the transmission of tool skills among themselves. It is unlikely a coincidence that both the ingenuity of roboticists and the iteration of evolution both converged on similar solutions; a novice cannot reliably acquire a new motor skill by merely observing an expert’s movements, novices must also peer into an expert’s mind.
早期灵长类动物进化出心智理论，用于政治活动。但这种能力被重新用于模仿学习。推断他人意图的能力使早期灵长类动物能够过滤掉无关的行为，只关注相关行为（这个人到底想做什么？）；它帮助年轻人长期专注于学习；它可能使早期灵长类动物能够通过推断新手会做什么和不懂什么来积极地互相教学。虽然我们的祖先哺乳动物可能可以通过观察他人来选择已知的技能，但正是在早期灵长类动物具备心智理论后，通过观察获得真正新颖的技能的能力才出现。这创造了一种新的可传递性：由聪明的个体发现的技能，曾经会随着他们去世而消失，现在可以在整个群体中传播，并一代又一代地传承下去。这就是为什么灵长类动物使用锤子而老鼠不使用的原因。
Theory of mind evolved in early primates for politicking. But this ability was repurposed for imitation learning. The ability to infer the intent of others enabled early primates to filter out extraneous behaviors and focus only on the relevant ones (what did the person mean to do?); it helped youngsters stay focused on learning over long stretches of time; and it may have enabled early primates to actively teach each other by inferring what a novice does and does not understand. While our ancestral mammal likely could select known skills by observing others, it was with early primates, armed with theory of mind, when the ability to acquire truly novel skills through observation emerged. This created a new degree of transmissibility: skills that were discovered by clever individuals and that would once have faded when they died, could now propagate throughout a group and be passed down endlessly through generations. This is why primates use hammers and rats do not.
18
18
为什么老鼠不能去杂货店购物
Why Rats Can’t Go Grocery Shopping
尽管罗宾·邓巴的社会大脑假说在过去的几十年里一直是科学家们对灵长类动物大脑扩张的主要解释，但还有另一种解释：所谓的生态大脑假说。
ALTHOUGH ROBIN DUNBAR’S social-brain hypothesis has, for the past several decades, held primacy among scientists as the leading explanation of brain expansion in primates, there is an alternative explanation: what has been called the ecological-brain hypothesis.
正如我们所见，早期灵长类动物不仅具有独特的社会性，而且饮食也很独特：它们是水果食性动物。以水果为主的饮食带来了一些令人惊讶的认知挑战。水果成熟后，只有一小段时间还没有落到森林的地面上。事实上，对于这些灵长类动物吃的许多水果来说，这段时间是少于七十二小时。有些树一年中只有不到三周的时间结出成熟的果实。有些水果很少有动物竞争对手（比如香蕉，因为果皮很难剥开），而其他水果则有很多动物竞争对手（比如无花果，任何动物都很容易吃到）。这些受欢迎的水果可能很快就会消失，因为一旦成熟，许多不同的动物都会以它们为食。总之，这意味着灵长类动物需要追踪大片森林中的所有水果，并且知道在任何一天哪种水果可能成熟；而成熟的水果可能最受欢迎，因此会最先消失。
As we have seen, early primates were not only uniquely social but also had a unique diet: they were frugivores. Fruit-based diets come with several surprising cognitive challenges. There is only a small window of time when fruit is ripe and has not yet fallen to the forest floor. In fact, for many of the fruits these primates ate, this window is less than seventy-two hours. Some trees offer ripe fruit for less than three weeks of the year. Some fruit has few animal competitors (such as bananas in their hard-to-open skin), while other fruit has many animal competitors (such as figs, which are easy for any animal to eat). These popular fruits are likely to disappear quickly, as many different animals feed on them once they ripen. Altogether this meant that primates needed to keep track of all the fruit in a large area of forest and on any given day know which fruit was likely to be ripe; and of the fruit that was ripe, which was likely to be most popular and hence disappear first.
研究表明，黑猩猩会计划夜间筑巢地点，为第二天的觅食做准备。对于无花果等更受欢迎的水果，它们会特意计划睡觉的地方，以便途中觅食。对于价格较低的水果，它们不会这样做竞争性但同样令人愉快的水果。此外，黑猩猩在追逐竞争性水果时会比在寻找其他水果时更早出发。竞争力较弱的水果。研究表明，狒狒也会提前计划它们的觅食旅程，当水果不太丰富时，它们会提早出发，而且很可能更快耗尽。
Studies have shown that chimpanzees plan their nighttime nesting locations in preparation for foraging on the subsequent day. For fruits that are more popular, such as figs, they will go out of their way to plan where they sleep to be en route to these fruits. They do not do the same for less competitive but equally enjoyable fruits. Further, chimpanzees will leave earlier in the morning when pursuing a competitive fruit than when traveling to forage for a less competitive fruit. It has been shown that baboons also plan their foraging journey in advance and will leave earlier when fruit is less abundant and likely to be depleted more quickly.
以非果类植物为食的动物不必应对同样的挑战；树叶、花蜜、种子、草和木材都能存活很长时间，而且不会分散在稀疏的斑块中。即使是食肉动物也没有那么具有挑战性的任务——必须猎捕猎物并智取胜，但很少只有很短的时间来捕猎。
Animals who feed on non-fruit plants do not have to cope with this same challenge; leaves, nectar, seeds, grass, and wood all last for long periods and are not localized in sparse patches. Even carnivores don’t have as cognitively challenging a task—prey must be hunted and outsmarted, but there are rarely only short time windows in which hunting is possible.
这种食果动物策略之所以如此具有挑战性，部分原因在于它不仅需要模拟不同的导航路径，还需要模拟你自己未来的需求。食肉动物和不吃水果的食草动物只有在饥饿时才能通过狩猎或放牧生存。但食果动物必须在饥饿之前提前计划行程。前一天晚上在前往附近受欢迎的果园的途中扎营需要预料到这样一个事实：如果你今晚不采取先发制人的措施早点找到食物，明天你就会挨饿。
Part of what makes this frugivore strategy so challenging is that it requires not only simulating differing navigational paths but also simulating your own future needs. Both a carnivore and a non-fruit-eating herbivore can survive by hunting or grazing only when they are hungry. But a frugivore must plan its trips in advance before it is hungry. Setting up camp en route to a nearby popular fruit patch the night before requires anticipating the fact that you will be hungry tomorrow if you don’t take preemptive steps tonight to get to the food early.
其他哺乳动物，比如老鼠，在冬季来临的时候，显然会囤积食物，在洞穴中储存大量坚果，以便在树木几乎不产出食物的漫长时期中生存下来。但这种季节性的囤积远不如每天根据明天的饥饿程度改变计划那么具有挑战性。此外，我们甚至不清楚老鼠囤积食物是否是因为它们知道自己将来会挨饿。事实上，实验室老鼠——尽管它们从未经历过没有食物的寒冷冬天——只要你降低它们环境的温度，它们就会自动开始囤积食物，这种现象只出现在北方物种的老鼠身上，它们必须进化是为了在冬天生存。因此，这似乎不是它们从过去的冬天学到并巧妙应对的活动；似乎这种囤积行为是对季节变化的进化硬编码反应。
Other mammals, such as mice, clearly stock up on food as winter months approach, storing vast reserves of nuts in their burrows to survive the long stretch when trees produce little to no food. But such seasonal hoarding is not nearly as cognitively challenging as the daily need to change your plans based on how hungry you will be tomorrow. Further, it isn’t even clear that mice hoard food because they understand that they will be hungry in the future. Indeed, lab mice—although they have never suffered from a cold winter without food—automatically start hoarding food if you simply lower the temperature of their environment, an effect seen only in northern species of mice who have had to evolve to survive winters. Therefore, this doesn’t seem to be an activity that they have learned from past winters and cleverly responded to; it seems that such hoarding is an evolutionarily hard-coded response to the changing seasons.
生态脑假说认为，早期灵长类动物的果食性饮食推动了它们大脑的快速扩张。2017 年，纽约大学的 Alex DeCasien 发表了一项研究，研究了超过140 种灵长类动物。有些灵长类动物主要以水果为食；其他灵长类动物现在主要以食叶动物为主（以树叶为食）。一些灵长类动物生活在非常小的社会群体中；而其他灵长类动物则生活在较大的社会群体中。他惊讶地发现，食果动物似乎比灵长类动物社会群体的大小更能解释相对大脑尺寸的变化。
The ecological-brain hypothesis argues that it was the frugivore diet of early primates that drove the rapid expansion of their brains. In 2017, Alex DeCasien from NYU published a study examining the diets and social lives of over 140 species of primates. Some primates are primarily frugivores; others are now primarily folivores (feeding on leaves). Some primates live in very small social groups; others live in large ones. He surprisingly found that being a frugivore seemed to explain the variation in relative brain size perhaps even better than the size of a primate’s social group.
比绍夫-科勒假说
The Bischof-Kohler Hypothesis
20 世纪 70 年代，两位比较心理学家，多丽丝·比绍夫-科勒 (Doris Bischof-Kohler) 和她的丈夫诺伯特·比绍夫 (Norbert Bischof)，提出了一个关于人类计划独特之处的新假设：他们假设，虽然其他动物可以根据当前需求制定计划（例如，当它们饿的时候如何获取食物），但只有人类可以根据未来需求制定计划（例如，即使你现在不饿，也要为下周的旅行获取食物）。进化心理学家托马斯·萨登多夫 (Thomas Suddendorf) 后来将此称为“比绍夫-科勒假设”。
In the 1970s, two comparative psychologists by the name of Doris Bischof-Kohler and her husband, Norbert Bischof, proposed a novel hypothesis about what was unique about planning in humans: They hypothesized that while other animals can make plans based on current needs (like how to get to food when they are hungry), only humans can make plans based on future needs (like how to get food for your trip next week, even though you are not hungry right now). The evolutionary psychologist Thomas Suddendorf would later call this the “Bischof-Kohler hypothesis.”
人类总是在预测未来的需求。即使我们不饿，我们也会去买东西；即使我们不冷，我们也会带上暖和的衣服去旅行。根据 Bischof-Kohler 时代的证据，只有人类才能做到这一点是一个合理的假设。但最近的证据对这一假设提出了质疑。现在有传闻说，当黑猩猩知道外面很冷时，它们会把温暖的笼子里的稻草带到外面筑巢，但在它们变冷之前。有研究发现，倭黑猩猩和猩猩会提前 14 个小时选择用于未来任务的工具这项任务。黑猩猩会从很远的地方搬石头去打开坚果，没有合适的石材，只能在一个地方制造工具在另一个地方使用。事实上，如果食果主义需要在你饿之前提前计划，那么我们应该期望灵长类动物能够预测未来的需求。
Humans anticipate future needs all the time. We go grocery shopping even when we are not hungry; we bring warm clothes on trips even when we are not cold. Given the evidence available at the time of Bischof-Kohler, it was a reasonable hypothesis that only humans could do this. But recent evidence has called this into question. There are now anecdotal stories of chimpanzees bringing straw from inside a warm cage to make a nest outside when they know it is cold outside but before they were cold. There have been findings of bonobos and orangutans selecting tools for future tasks up to fourteen hours in advance of that task. Chimpanzees will carry stones from faraway locations to open nuts in areas that have no suitable stones and will manufacture tools in one location for use in another location. Indeed, if frugivorism requires planning ahead before you are hungry, then we should expect primates to be able to anticipate future needs.
2006 年，西安大略大学的 Miriam Naqshbandi 和 William Roberts 测量了一只松鼠猴和一只老鼠预测自己未来口渴的能力，并相应地改变它们的行为。研究人员给松鼠猴和老鼠两个杯子，让它们选择其中一种。杯子 1 是“小份零食”，里面装着一小块食物，杯子 2 是“大份零食”，里面装着很多食物。松鼠猴的零食是枣子，老鼠的零食是葡萄干。在正常情况下，两种动物都会选择高热量的食物；它们喜欢枣和葡萄干。
In 2006, Miriam Naqshbandi and William Roberts of the University of Western Ontario measured a squirrel monkey’s and a rat’s ability to anticipate their own future thirst and change their behavior accordingly. Squirrel monkeys and rats were both given two options by being presented with two cups. Cup 1 was a “small treat option,” which contained a tiny morsel of food, and cup 2 was a “high treat option,” which contained lots of food. For squirrel monkeys the treats were dates; for rats they were raisins. Under normal conditions, both animals would choose the high treat option; they love dates and raisins.
但纳克什班迪和罗伯茨随后在不同的条件下对这些动物进行了测试。枣和葡萄干会诱发这些动物的大量口渴，通常需要它们喝两倍以上的水来补充水分。那么，如果这些动物被迫做出权衡，考虑到它们未来的口渴状态，会发生什么？纳克什班迪和罗伯茨修改了测试，如果动物选择高奖励选项（杯子里有很多枣或葡萄干），它们要几个小时后才能喝到水；然而，如果动物选择低奖励选项（杯子里只有少量枣或葡萄干），它们在 15 到 30 分钟后才能喝到水。会发生什么？
But Naqshbandi and Roberts then tested these animals in a different condition. Dates and raisins induce large amounts of thirst in these animals, often requiring them to consume over twice as much water to rehydrate themselves. So what happens if these animals are forced to make a trade-off, incorporating their future state of thirst? Naqshbani and Roberts modified the test such that if animals select the high treat option (the cup with many dates or raisins), they will only get access to water hours later; however, if animals select the low treat option (the cup with few dates or raisins) they get access to water between 15 and 30 minutes later. What happens?
有趣的是，松鼠猴学会了选择低奖励选项，而老鼠则继续选择高奖励选项。松鼠猴能够抵制现在吃零食的诱惑，因为它们期待着某种东西——水——它们甚至还不想要。换句话说，猴子可以做出决定，以预期未来的需要。相比之下，老鼠完全无法做到这一点——它们坚持错误的逻辑：“为什么要放弃额外的葡萄干来喝水，我甚至不渴！” *
Fascinatingly, squirrel monkeys learn to select the low treat option, while rats continue to select the high treat option. Squirrel monkeys are capable of resisting the temptation to have treats now, in anticipation of something—water—that they don’t even want yet. In other words, monkeys can make a decision in anticipation of a future need. In contrast, rats were entirely unable to do this—they stuck with the flawed logic of “why give up extra raisins for water, I’m not even thirsty!”*
这表明，萨登多夫的 Bischof-Kohler 假说可能是正确的，即预测未来的需求是一种更困难的规划形式，并且某些动物应该能够规划但无法预测未来的需求（例如老鼠）。但可能并非只有人类才拥有这种能力。相反，这可能是许多灵长类动物的领域。
This suggests that perhaps Suddendorf’s Bischof-Kohler hypothesis was correct that anticipating a future need is a more difficult form of planning and was correct that some animals should be able to plan but unable to anticipate future needs (such as rats). But it may not be the case that only humans were endowed with this ability. It may instead be the province of many primates.
灵长类动物如何预测未来的需求
How Primates Anticipate Future Needs
基于预期需求（你目前没有遇到的需求）做出选择的机制，对老年人来说是一个困境哺乳动物的大脑结构。我们推测大脑皮层控制行为的机制是通过间接模拟决策，然后由较老的脊椎动物结构（基底神经节、杏仁核和下丘脑）评估决策的结果。这种机制允许动物只选择模拟的路径和行为，这些路径和行为会立即激发正价神经元，例如饥饿时想象食物或口渴时想象水。
The mechanics of making a choice based on an anticipated need, one you are not currently experiencing, presents a predicament to the older mammalian brain structures. We have speculated that the mechanism by which the neocortex controls behavior is by simulating decisions vicariously, the outcomes of which are then evaluated by the older vertebrate structures (basal ganglia, amygdala, and hypothalamus). This mechanism allows an animal to choose only simulated paths and behaviors that excite positive valence neurons right now, like imagining food when hungry or water when thirsty.
相比之下，为了购买一周的食品杂货，我需要预计披萨将成为周四电影之夜的绝佳补充，即使我目前并不想吃披萨。当我想象在不饿的情况下吃披萨时，我的基底神经节不会兴奋；它不会为任何吃披萨的决定积累投票。因此，为了想吃披萨，我需要意识到，在这种想象的未来饥饿状态下，食物的气味和外观会刺激正价神经元，即使现在想象它不会。那么，在没有任何替代正价激活的情况下，大脑如何选择想象的路径？你的大脑皮层怎么会想要杏仁核和下丘脑不想要的东西呢？
In contrast, to buy groceries for the week, I need to anticipate a pizza is going to make a great addition to Thursday’s movie night, even though I don’t currently want pizza. When I imagine eating pizza while I’m not hungry, my basal ganglia doesn’t get excited; it doesn’t accumulate votes for any decisions to pursue pizza. Thus, to want pizza, I need to realize that in this imagined future state of hunger, the smell and sight of food will excite positive valence neurons, even though imagining it right now does not. How, then, can a brain choose an imagined path in the absence of any vicarious positive-valence activation? How can your neocortex want something that your amygdala and hypothalamus do not?
我们已经看到过另一种情况，大脑需要推断出一种意图，即“想要”，而目前大脑还没有这种意图：当大脑试图推断其他人的需要时。大脑是否能够使用相同的心理理论机制来预测未来的需求？换句话说：想象别人的想法和想象未来自己的想法真的有什么不同吗？
There is another situation we have already seen where brains need to infer an intent—a “want”—of which it does not currently share: when they’re trying to infer the wants of other people. Might brains be able to use the same mechanism of theory of mind to anticipate a future need? Put another way: Is imagining the mind of someone else really any different from imagining the mind of your future self?
也许我们预测未来需求的机制与我们进行心智理论的机制相同：我们可以推断出心智的意图——无论是我们自己的还是别人的——在与当前情况不同的其他情况下。就像我们可以正确地推断出一个被剥夺食物的人的渴望（“如果詹姆斯 24 小时不吃东西，他会有多饿？”），即使我们自己可能不饿，也许我们也可以推断出我们自己在未来情况下的意图（“如果我 24 小时不吃东西，我会有多饿？”），即使我们目前不饿。
Perhaps the mechanism by which we anticipate future needs is the same mechanism by which we engage in theory of mind: We can infer the intent of a mind—whether our own or someone else’s—in some different situation from our current one. Just as we can correctly infer the cravings of someone deprived of food (“How hungry would James be if he didn’t eat for twenty-four hours?”) even though we ourselves might not be hungry, perhaps too we can infer the intent of ourselves in a future situation (“How hungry would I be if I didn’t eat for twenty-four hours?”) even though we are currently not hungry.
托马斯·萨登多夫 (Thomas Suddendorf) 在讨论比绍夫-科勒假说的论文中精辟地预示了这一观点：
In his paper discussing the Bischof-Kohler hypothesis, Thomas Suddendorf brilliantly foreshadowed exactly this idea:
图 18.1：心智理论与预测未来需求之间的相似性
Figure 18.1: The similarity between theory of mind and anticipating future needs
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
未来需求预期……可能只是动物普遍存在的一个特殊问题，即同时表现出相互冲突的心理状态。就像 3 岁的孩子一样，它们可能无法想象与现在不同的早期信念（或知识状态、驱动力等），也无法理解另一个人持有与自己不同的信念。这可能适用于未来状态，也适用于过去状态。也就是说，饱腹的动物可能无法理解自己以后可能会饿，因此可能无法采取措施确保未来饥饿就会得到满足。
Future need anticipation . . . might be only a special case of animals’ general problem with simultaneously representing conflicting mental states. Like 3-year-old children, they may be unable to imagine an earlier belief (or state of knowledge, or drive, etc.) that is different from a present one or to understand that another individual holds a belief different from their own. This may apply to future states as well as to past ones. That is, a satiated animal may be unable to understand that it may later be hungry, and therefore may be unable to take steps to ensure that this future hunger will be satisfied.
尽管纳格什班迪和罗伯茨对松鼠猴和老鼠的实验表明，萨登多夫关于只有人类才能预测未来需求的观点可能是错误的，但萨登多夫可能很有先见之明地提出，模拟与现实生活分离的心理状态的一般能力你自己可以重新用于心智理论并预测未来的需求。
Although Naqshbandi and Roberts’s experiment with squirrel monkeys and rats suggests that Suddendorf may have been wrong that only humans can anticipate future needs, Suddendorf may have been prescient in proposing that the general ability to model a dissociated mental state from your own can be repurposed for both theory of mind and anticipating future needs.
有两个观察结果支持这一观点。首先，似乎心智理论和预测未来需求的能力都存在于灵长类动物中，甚至以原始形式存在，但在许多其他哺乳动物中却不存在，这表明这两种能力在早期灵长类动物中大约同时出现。其次，人们在心智理论和预测未来需求的任务中会犯类似类型的错误。
There are two observations that support this idea. First, it seems that both theory of mind and anticipating future needs are present, even in a primitive form, in primates, but not in many other mammals, suggesting both abilities emerged around the same time in early primates. Second, people make similar types of mistakes in tasks of theory of mind and of anticipating future needs.
例如，我们在第 16 章中看到，口渴的人会错误地预测其他人也一定口渴。同样，饥饿的人似乎也会错误地预测他们将来需要多少食物。带两组人去杂货店为自己购买一周的食物，饥饿的人最终会比其他人购买更多的食物吃得好，尽管他们都在购买食物，以满足同一时间段（即一周）的需要。当你饿的时候，你会高估自己未来的饥饿程度。
For example, we saw in chapter 16 that thirsty people become incorrectly biased to predict that other people must also be thirsty. Well, it is also the case that hungry people seem to incorrectly predict how much food they will need in the future. Take two groups of people and bring them to the grocery store to shop for themselves for the week, and those that are hungry will end up buying more food than those that were well fed, even though they were both shopping for food to feed themselves for the same window of time, namely, a single week. When hungry, you overestimate your own future hunger.
预测未来需求的能力会给我们祖先的果食动物带来诸多好处。它使我们的祖先能够提前很久规划他们的觅食路线，从而确保他们能最先采摘到刚成熟的果实。我们今天为遥远、抽象和尚不存在的目标做出决策的能力是从树栖灵长类动物那里继承来的。这一技巧最初可能是为了最先采摘果实而使用的，但如今，在人类中，它被用于更伟大的目的。它为我们在很长一段时间内制定长期计划的能力奠定了基础。
The ability to anticipate future needs would have offered numerous benefits to our ancestral frugivores. It would have enabled our ancestors to plan their foraging routes long in advance, thereby ensuring they were the first to get newly ripened fruits. Our ability to make decisions today for faraway, abstract, and not-yet-existent goals was inherited from tree faring primates. A trick that, perhaps, was first used for getting the first pick of fruits, but today, in humans, is used for far greater purposes. It laid the foundation for our ability to make long term plans over vast stretches of time.
突破#4：心理化
Summary of Breakthrough #4: Mentalizing
早期灵长类动物似乎已具备三大能力：
There are three broad abilities that seem to have emerged in early primates:
- 心智理论：推断他人的意图和知识
- Theory of mind: inferring intent and knowledge of others
- 模仿学习：通过观察获得新技能
- Imitation learning: acquiring novel skills through observation
- 预测未来的需求：现在采取行动来满足未来的需要，即使我现在不想要
- Anticipating future needs: taking an action now to satisfy a want in the future, even though I do not want it now
事实上，这些可能不是单独的能力，而是单一新突破的涌现属性：构建自己的思维生成模型，这一技巧可以称为“心理化”。我们看到这些能力源自共同的神经结构（如 gPFC），这些结构首先在早期灵长类动物中进化而来。我们看到孩子们似乎获得了这些能力在相似的发育时期。我们看到，损害其中一种能力的损伤往往会损害其中多种能力。
These may not, in fact, have been separate abilities but rather emergent properties of a single new breakthrough: the construction of a generative model of one’s own mind, a trick that can be called “mentalizing.” We see this in the fact that these abilities emerge from shared neural structures (such as the gPFC) that evolved first in early primates. We see this in the fact that children seem to acquire these abilities at similar developmental times. We see this in the fact that damage that impairs one of these abilities tends to impair many of them.
最重要的是，我们看到，这些技能产生的结构与我们推理自己思维的能力产生的区域是相同的。这些新的灵长类动物区域不仅需要模拟他人的思维，还需要将自己投射到想象的未来、在镜子中识别自己（镜像符号综合症）和识别自己的动作（异手综合症）。而孩子推理自己思维的能力往往先于孩子这三种能力的发展。
And most important, we see this in the fact that the structures from which these skills emerge are the same areas from which our ability to reason about our own mind emerges. These new primate areas are required not only for simulating the mind of others but also for projecting yourself into your imagined futures, identifying yourself in the mirror (mirror-sign syndrome), and identifying your own movements (alien-hand syndrome). And a child’s ability to reason about her own mind tends to precede a child’s development of all three of these abilities.
然而，这一想法的最佳证据可以追溯到蒙特卡斯尔。早期灵长类动物的大脑除了大小之外，主要的变化是增加了新的大脑皮层区域。因此，如果我们坚持由蒙特卡斯尔、亥姆霍兹、辛顿、霍金斯、弗里斯顿和其他许多人启发的一般观点，即大脑皮层的每个区域都由相同的微电路组成，那么这将对我们解释灵长类动物新发现的能力施加严格的限制。这表明，这些新的智力技能必须来自大脑皮层的一些新的巧妙应用而不是一些新颖的计算技巧。这使得将心智理论、模仿学习和对未来需求的预期解释为二阶生成模型的涌现属性成为一个很好的提议——这三种能力都可能源自新皮层的新应用。
However, the best evidence for this idea goes all the way back to Mountcastle. The main change to the brain of early primates, besides its size, was the addition of new areas of neocortex. So if we are to stick to the general idea—inspired by Mountcastle, Helmholtz, Hinton, Hawkins, Friston, and many others—that every area of neocortex is made up of identical microcircuits, this imposes strict constraints on how we explain the newfound abilities of primates. It suggests that these new intellectual skills must emerge from some new clever application of the neocortex and not some novel computational trick. This makes the interpretation of theory of mind, imitation learning, and anticipation of future needs as nothing more than an emergent property of a second-order generative model a nice proposal—all three abilities can emerge from nothing more than new applications of neocortex.
图 18.2
Figure 18.2
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
所有这些能力——心智理论、模仿学习和预测未来需求——在早期灵长类动物的独特环境中都具有特别的适应性。邓巴认为，社会脑假说和生态脑假说是同一枚硬币的两面。心智化能力可能同时释放了成功采摘水果和成功政治的能力。食果主义和社会等级的压力可能汇聚在一起，产生了持续的进化压力，促使人们开发和完善大脑区域——例如 gPFC——以塑造自己的思维。
All these abilities—theory of mind, imitation learning, and anticipating future needs—would have been particularly adaptive in the unique niche of early primates. Dunbar argues that the social-brain hypothesis and the ecological-brain hypothesis are two sides of the same coin. The ability to mentalize may have simultaneously unlocked both the ability to successfully forage fruits and to successfully politick. The pressures of both frugivorism and social hierarchies may have converged to produce continual evolutionary pressure to develop and elaborate brain regions—such as the gPFC—for modeling your own mind.
我们已经到达了第四个突破的尽头。在我们故事的这个节点，我们正站在人类和我们现存最亲近的亲属之间最后分道扬镳的边缘。我们与黑猩猩的共同祖先生活了七百万年大约公元前 1500 年前，黑猩猩出现在东非。这个祖先的后代分化出两条进化路径，一条进化成了今天的黑猩猩，另一条进化成了今天的人类。
We have reached the end of breakthrough #4. At this point in our story, we now stand at the precipice of the final divergence between humankind and our closest living relative. Our shared ancestor with chimpanzees lived seven million years ago in eastern Africa. The offspring of this ancestor split down two evolutionary paths—one that became today’s chimpanzees, and another that became today’s humans.
如果我们把六亿年的进化时间（从第一个大脑出现到今天）压缩成一个那么我们现在就处于圣诞夜，也就是十二月的最后七天。在接下来的“七天”里，我们的祖先将从采摘水果到驾驶猎鹰 9 号火箭。让我们来看看具体是怎么回事。
If we were to scrunch the six hundred million years of evolutionary time—from which the first brains emerged until today—into a single calendar year, then we would now find ourselves perched at Christmas Eve, the final seven days of December. Over the next “seven days,” our ancestors will go from foraging fruits to flying Falcon 9 rockets. Let’s find out how.
突破＃5
Breakthrough #5
说话和第一批人类
Speaking and the First Humans
10万年前的大脑
Your brain 100,000 years ago
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
19
19
探索人类的独特性
The Search for Human Uniqueness
几千年来，我们人类一直带着自鸣得意的自豪感看着镜子中的自己，思考着我们与其他动物相比的诸多优越之处。亚里士多德声称，人类独有的“理性灵魂”——推理、抽象和反思的能力。20 世纪的动物心理学家列举了许多他们认为是人类独有的智力能力。一些人认为只有人类才能进行心理时间旅行。另一些人认为，这是我们的情景记忆。还有一些人认为，这是我们预测未来需求的能力。还有一些人认为，这是我们的自我意识。还有一些人认为，这是我们沟通、协调和使用工具的能力。诸如此类。
FOR MILLENNIA, WE HUMANS have looked at ourselves in the mirror with self-congratulatory pride and pondered the many ways in which we are superior to our fellow animals. Aristotle claimed it was our “rational soul”—our ability to reason, make abstractions, and reflect—that was uniquely human. Twentieth-century animal psychologists itemized many intellectual abilities they believed were uniquely human. Some argued that only humans engage in mental time travel. Others, that it is our episodic memory. Others, our ability to anticipate future needs. Others, our sense of self. Our ability to communicate, coordinate, use of tools. The lists went on.
但上个世纪对其他动物行为的研究已经系统地拆解了我们这个脆弱得令人惊讶的独特性大厦。尽管直觉上认为这些技能中有许多是人类独有的，但正如我们在本书中看到的那样，科学表明，其中许多技能（如果不是全部的话）可能根本不是人类独有的。
But the last century of research into the behaviors of other animals has methodically dismantled our surprisingly fragile edifice of uniqueness. Despite the intuitive appeal of claiming many of these skills as only ours, as we have seen throughout this book, science suggests many of them, if not all, might not be unique to humans at all.
达尔文认为，“人类与高等动物在心智上的差异，尽管很大，但肯定只是程度上的差异”。而不是同类。” 哪些智力成就（如果有的话）是人类独有的，心理学家们仍在激烈争论。但随着证据不断涌现，达尔文似乎是正确的。
Darwin believed that “the difference in mind between man and the higher animals, great as it is, is certainly one of degree and not of kind.” What intellectual feats, if any, are uniquely human is still hotly debated among psychologists. But as the evidence continues to roll in, it seems that Darwin may have been right.
如果人类拥有众多独一无二的智力能力，我们就会期望人类的大脑包含一些独特的神经结构、一些新的线路、一些新的系统。但证据却恰恰相反——人类大脑中没有一种神经结构不是在我们的同类猿类的大脑中发现的，有证据表明，人类大脑实际上只是一个放大的灵长类大脑：一个更大的新皮层，一个更大的基底神经节，但仍然包含所有相同的连接区域以同样的方式。扩大黑猩猩的大脑规模可能让我们能够更好地预测未来的需求、心智理论、运动技能和规划，但它并不一定能给我们带来任何真正新的东西。
If it were the case that humans wielded numerous intellectual capabilities that were entirely unique in kind, we would expect human brains to contain some unique neurological structures, some new wiring, some new systems. But the evidence is the opposite—there is no neurological structure found in the human brain that is not also found in the brain of our fellow apes, and evidence suggests that the human brain is literally just a scaled-up primate brain: a bigger neocortex, a bigger basal ganglia, but still containing all the same areas wired in all the same ways. Scaling up a chimpanzee brain may have made us better at anticipating future needs, theory of mind, motor skills, and planning, but it didn’t necessarily give us anything truly new.
自从人类与黑猩猩分化以来，对人类大脑进化的一个非常合理的解释是，各种进化压力导致我们的人类祖先只是“提升”了已经存在的能力。
An eminently reasonable explanation of human brain evolution since our divergence with chimpanzees is that various evolutionary pressures led our human ancestors to simply “level up” the abilities that were already present.
难道根本就没有突破吗？
Was there perhaps no breakthrough at all?
这似乎是最合理的解释——但有一个关键的例外。正是在这个例外中，我们第一次看到了人性的意义。
This seems to be the most reasonable interpretation—but for one crucial exception. And it is in this singular exception that we see the first hints of what it means to be human.
我们独特的沟通方式
Our Unique Communication
早在早期人类说出第一句话之前，生物体就已经开始相互交流。单细胞生物会发出化学信号来分享基因和环境信息。无脑海葵会向水中释放信息素，以协调精子和卵子释放的时间。蜜蜂会跳舞来发出寻找食物的信号。鱼类会用电信号相互求爱。爬行动物会点头来传达攻击性。老鼠会发出吱吱声来表达危险或兴奋。生物之间的交流在进化史上是古老而普遍的。
Organisms had been communicating with each other long before early humans uttered their first words. Single-celled organisms emit chemical signals to share genes and information about the environment. Brainless sea anemones pump pheromones into the water to coordinate the timing of sperm and egg release. Bees dance to signal where to find food. Fish use electrical signals to court each other. Reptiles head-bob to communicate their aggressiveness. Mice squeak to express danger or excitement. Communication between organisms is evolutionarily ancient and ubiquitous.
黑猩猩不停地尖叫，互相打手势。这些不同的声音和手势已被证明可以传达特定的请求。拍肩表示“停止”，跺脚表示“和我一起玩”，尖叫表示“给我梳毛”，伸出手掌表示“分享食物”。灵长类动物学家对这些不同的手势和发声进行了广泛的研究，甚至有一本大猿词典记录了近一百种声音和手势。
Chimpanzees are constantly screeching and gesturing to each other. These different sounds and gestures have been shown to signal specific requests. Shoulder taps mean “Stop it,” stomping means “Play with me,” a screech means “Groom me,” reaching a palm out means “Share food.” Primatologists have studied these different gestures and vocalizations so extensively that there is even a Great Ape Dictionary that has chronicled almost one hundred sounds and gestures.
长尾黑尾猴会发出不同的声音来表示特定捕食者的存在。当一只猴子发出“豹子！”的尖叫声时，其他猴子都会跑到树上。当一只猴子发出“老鹰！”的尖叫声时，其他猴子都会跳到森林地面上。实验人员只需从附近的扬声器播放其中一种声音，就可以让所有猴子跑到树梢或跳到地面上。
Vervet monkeys have different sounds to signal the presence of specific predators. When one monkey makes the squeal meaning “Leopard!” all the others run to the trees. When one makes the squeal meaning “Eagle!” all the others jump to the forest floor. Experimenters can get all the monkeys to run to the treetops or jump to the floor simply by playing one of these sounds from nearby loudspeakers.
当然，还有我们——智人。我们也会相互交流。我们的独特之处并不在于我们交流，而是在于我们交流的方式。人类使用语言。
And then, of course, there is us—Homo sapiens. We too communicate with each other. It is not the fact that we communicate that is unique; rather, it is how we communicate. Humans use language.
人类语言与其他动物交流方式有两点不同。首先，已知的其他自然动物交流方式都不会赋予陈述性标签（也称为符号）。人类老师会指向某个物体或行为并赋予其任意标签：大象、树、奔跑。与此相反，其他动物的交流方式是遗传固定的，并未指定。在没有相互接触的不同群体中，长尾黑猩猩和黑猩猩的手势几乎完全相同。缺乏社会接触的猴子和类人猿仍然使用相同的手势。事实上，这些手势甚至在灵长类动物之间也是共享的；倭黑猩猩和黑猩猩几乎拥有完全相同的手势和发声方式。在非人类灵长类动物中，这些手势和发声的含义不是通过陈述性标签赋予的，而是直接来自遗传固定。
Human language differs from other forms of animal communication in two ways. First, no other known form of naturally occurring animal communication assigns declarative labels (otherwise known as symbols). A human teacher will point to an object or a behavior and assign it an arbitrary label: elephant, tree, running. In contrast, other animals’ communications are genetically hardwired and not assigned. Vervet monkey and chimpanzee gestures are almost identical across different groups that have no contact with each other. Monkeys and apes deprived of social contact still use the same gestures. In fact, these gestures are even shared across species of primates; bonobos and chimpanzees share almost exactly the same repertoire of gestures and vocalizations. In nonhuman primates, the meanings of these gestures and vocalizations is not assigned through declarative labeling but emerge directly from genetic hardwiring.
那么教狗或其他动物命令怎么样？显然，这代表了某种形式的标记。语言学家区分了陈述性标签和命令性标签。命令性标签是能产生奖励的标签：“当我听到“坐下”时，如果我坐下，我就会得到一份零食”或“当我听到“待”时，如果我停止移动，我就会得到一份零食。”这是基本的时间差异学习 - 所有脊椎动物都能做到这一点。另一方面，陈述性标记是人类语言的一个特殊特征。陈述性标签是为对象或行为分配任意符号的标签 - “那是一头牛”，“那是在跑” - 完全没有任何命令性。尚未发现其他形式的自然发生的动物交流能够做到这一点。
What about teaching a dog or any other animal a command? Clearly this represents some form of labeling. Linguists make a distinction between declarative and imperative labels. An imperative label is one that yields a reward: “When I hear sit, if I sit, I will get a treat” or “When I hear stay, if I stop moving, I will get a treat.” This is basic temporal difference learning—all vertebrates can do this. Declarative labeling, on the other hand, is a special feature of human language. A declarative label is one that assigns an object or behavior an arbitrary symbol—“That is a cow,” “That is running,”—without any imperative at all. No other form of naturally occurring animal communication has been found to do this.
人类语言与其他动物交流方式的第二个不同之处在于，人类语言包含语法。人类语言包含规则，我们通过这些规则合并和修改符号来传达特定的含义。我们可以把这些陈述性标签编织成句子，再把这些句子编织成概念和故事。这使我们能够将典型人类语言中存在的几千个单词转换成看似无限数量的独特含义。
The second way in which human language differs from other animal communication is that it contains grammar. Human language contains rules by which we merge and modify symbols to convey specific meanings. We can thereby weave these declarative labels into sentences, and we can knit these sentences into concepts and stories. This allows us to convert the few thousand words present in a typical human language into a seemingly infinite number of unique meanings.
语法最简单的方面是，我们说出符号的顺序传达了含义：“本拥抱了詹姆斯”和“詹姆斯拥抱了本”的意思不同。我们还嵌入了对顺序敏感的子短语：“伤心的本拥抱了詹姆斯”和“伤心的本拥抱了詹姆斯”的意思完全不同。但语法规则不仅仅是顺序。我们用不同的时态来传达时间：“本在攻击我”和“马克斯攻击了我”。我们有不同的冠词：“宠物叫了”和“宠物叫了”的意思不同。
The simplest aspect of grammar is that the order in which we utter symbols conveys meaning: “Ben hugged James” means something different from “James hugged Ben.” We also embed subphrases that are sensitive to ordering: “Ben, who was sad, hugged James” means something entirely different than “Ben hugged James, who was sad.” But the rules of grammar go beyond just order. We have different tenses to convey timing: “Ben is attacking me” versus “Max attacked me.” We have different articles: “The pet barked” means something different than “A pet barked.”
当然，这只是英语；地球上有六千多种口语语言，每种语言都有自己的标签和语法。但是，尽管不同语言的具体标签和语法千差万别，但人类发现的每一个群体都使用过语言。即使是澳大利亚和非洲的狩猎采集社会，在他们被“发现”时，已经与其他人类群体五万多年没有接触过，他们仍然说着自己的语言，这些语言和其他人类的语言一样复杂。这是无可辩驳的证据，表明人类的共同祖先说着自己的语言，有自己的陈述标签和语法。
And of course, this is just English; there are over six thousand spoken languages on Earth, each with its own labels and grammar. But despite the great diversity of the specific labels and grammars of different languages, every single group of humans ever discovered used language. Even hunter-gatherer societies in Australia and Africa who at the point they were “discovered” had had no contact with any other groups of humans for over fifty thousand years, still spoke their own languages that were equally complex as those of other humans. This is irrefutable evidence that the shared ancestor of humans spoke their own languages, with their own declarative labels and grammars.
当然，早期人类用自己的语言说话，并带有陈述性标签和语法，而其他动物却天生不具备这种能力，这一事实并不能证明只有人类才能使用语言，而只能证明只有人类碰巧使用语言。早期人类的大脑真的进化出了某种独特的说话能力吗？还是语言只是五万多年前发现的一种文化技巧，并简单地在所有现代人类中代代相传？语言是进化的发明还是文化的发明？
Of course, the fact that early humans spoke in their own languages with declarative labels and grammar, while no other animal naturally does so, does not prove that only humans are able to use language, merely that only humans happen to use language. Did the brains of early humans really evolve some unique ability to speak? Or is language just a cultural trick that was discovered over fifty thousand years ago and was simply passed down through generations of all modern humans? Is language an evolutionary invention or a cultural invention?
这里有一种方法可以验证这一点：如果我们试图教我们在进化上最亲近的动物表亲，也就是我们的同类猿类，会发生什么？如果猿类成功学习了语言，这表明语言是一种文化发明；如果猿类失败了，这表明它们的大脑缺乏人类大脑中出现的关键进化创新。
Here’s one way to test this: What happens if we try to teach language to our evolutionarily closest animal cousins, our fellow apes? If apes succeed in learning language, that suggests that language was a cultural invention; if apes fail, this suggests their brains lack a key evolutionary innovation that emerged in humans.
这项测试已进行多次。结果既令人惊讶又发人深省。
This test has been performed multiple times. The result is as surprising as it is revealing.
教授猿类语言的尝试
Attempts to Teach Apes Language
首先：我们无法真正教会猿类说话。20 世纪 30 年代曾有人尝试过，但失败了——非人类猿类在生理上无法产生口头语言。人类的声带是专门为说话而生的；人类的喉部较低，颈部较长，这使我们能够发出比其他猿类更广泛的元音和辅音。黑猩猩的声带只能发出有限的喘息和尖叫声。
To start: We can’t literally teach apes to speak. This was attempted in the 1930s, and it failed—nonhuman apes are physically incapable of producing verbal language. Human vocal cords are uniquely adapted to speech; the human larynx is lower and the human neck is longer, which enables us to produce a much wider variety of vowels and consonants than other apes. The vocal cords of a chimp can produce only a limited repertoire of huffs and squeals.
然而，语言之所以成为语言，并非因为媒介，而是因为实质——人类语言的许多形式都是非语言的。没有人会因为文字、手语和盲文不涉及发声而声称它们不包含语言的实质。
However, what makes language language is not the medium but the substance—many forms of human language are nonverbal. No one would claim that writing, sign language, and Braille do not contain the substance of language because they don’t involve vocalization.
试图教会黑猩猩、大猩猩和倭黑猩猩语言的关键研究使用了美国手语或虚构的视觉语言，其中猿类指向板上的符号序列。从婴儿时期开始，这些猿类就被训练使用这些语言，人类实验者一遍又一遍地用手势或指向符号来指代物体（苹果、香蕉）或动作（挠痒痒、玩耍、追逐），直到猿类开始重复这些符号。
The key studies that attempted to teach chimpanzees, gorillas, and bonobos language used either American Sign Language or made-up visual languages in which apes pointed to sequences of symbols on a board. Beginning as infants, these apes were trained to use these languages, with human experimenters signing or pointing to symbols to refer to objects (apples, bananas) or actions (tickling, playing, chasing) over and over again until the apes began to repeat the symbols.
在大多数研究中，经过多年的训练，非人类猿类确实能够做出相应的手势。它们看到狗时会做出“狗”的手势，看到鞋子时会做出“鞋子”的手势。
Across most of these studies, after years of being taught, nonhuman apes did indeed produce the appropriate signs. They could look at a dog and sign dog and look at a shoe and sign shoe.
它们甚至可以构造基本的名词-动词对。常用短语有play me和tickle me。一些证据甚至表明，它们可以将已知单词组合起来，创造出新的含义。在一个著名的轶事中，黑猩猩 Washoe 第一次看到一只天鹅，当训练员打手势说“那是什么？”时，Washoe 回答说“水鸟”。在另一个轶事中，大猩猩 Koko 看到了一枚戒指，不知道该怎么称呼它，签名，手指手镯。第一次吃羽衣甘蓝后，倭黑猩猩 Kanzi 按下了“慢生菜”的符号。
They could even construct basic noun-verb pairs. Common phrases were play me and tickle me. Some evidence even suggested they could combine known words to create novel meanings. In one famous anecdote, the chimpanzee Washoe saw a swan for the first time, and when the trainer signed, What’s that?, Washoe signed back, Water bird. In another, the gorilla Koko saw a ring and, not knowing the word for it, signed, Finger bracelet. After eating kale for the first time, Kanzi the bonobo pressed symbols for slow lettuce.
据称 Kanzi 甚至会用语言与他人玩耍。有个轶事说，一个训练员在 Kanzi 的栖息地休息时，被 Kanzi 抢走毯子，然后兴奋地按下“坏惊喜”符号吵醒。在另一个轶事中，Kanzi 按下了“追逐苹果”的键，然后拿起一个苹果，咧嘴一笑，开始逃离他的教练。
Kanzi supposedly even used language to play with others. There is an anecdotal story of a trainer who was resting in Kanzi’s habitat being woken up by Kanzi snatching away the blanket and then excitedly pressing the symbols for bad surprise. In another anecdote, Kanzi pressed keys for apple chase and then picked up an apple, grinned, and began running from his trainer.
苏·萨维奇-朗博是心理学家和灵长类动物学家，她设计了坎兹语言学习实验，并进行了一项测试，将坎兹的语言理解能力与两岁人类儿童的语言理解能力进行比较。萨维奇-朗博让坎兹和一个人类儿童接触了六百多个带有特定命令的新句子（使用他们的符号语言）。这些句子使用了坎兹已经知道的符号，但坎兹从未见过的句子，命令包括“你能把黄油给罗斯吗？”；“去给丽兹涂点肥皂；去拿冰箱里的香蕉；你能给小狗一个拥抱吗？”；“戴上怪物面具吓唬琳达”。坎兹成功地完成了这些任务。70% 的时间，其表现甚至超过了两岁的人类儿童。
Sue Savage-Rumbaugh, the psychologist and primatologist who devised the Kanzi language-learning experiment, did a test to compare Kanzi’s language understanding to that of a two-year-old human child. Savage-Rumbaugh exposed Kanzi and a human child to over six hundred novel sentences (using their symbol-language) with specific commands. These sentences used symbols Kanzi already knew but in sentences Kanzi had never seen, commands like Can you give the butter to Rose?; Go put some soap on Liz; Go get the banana that’s in the refrigerator; Can you give the doggy a hug?; and Put on the monster mask and scare Linda. Kanzi successfully completed these tasks over 70 percent of the time, outperforming the two-year-old human child.
这些猿类语言研究在多大程度上展示了带有陈述性标签和语法的语言，语言学家、灵长类动物学家和比较心理学家仍然存在争议。许多人认为，这些技巧代表的是命令式而不是陈述式，而且说出的短语非常简单，很难称为语法。事实上，在大多数这些研究中，猿类在使用正确的标签时会得到奖励，这使得很难判断它们是真的在分享标记物体，还是只是学会了如果它们在看到香蕉时确实用手势 X，它们就会得到奖励，这是任何无模型强化学习机器都可以执行的任务。对这些能够使用语言的猿类说出的短语进行广泛分析表明，短语的多样性较低，这意味着它们倾向于使用它们学到的确切短语（例如，Tickle me），而不是将单词组合成新的短语（例如，我想被挠痒痒）。但针对这些挑战，许多人指出了萨维奇-朗博的研究和坎齐对命令和俏皮短语的极其准确的语法理解。争论尚未结束。
The degree to which these ape language studies demonstrate language with declarative labels and grammar is still controversial among linguists, primatologists, and comparative psychologists. There are many who argue that these tricks represent imperatives not declaratives and that the phrases uttered were so simple they could hardly be called grammar. Indeed, in most of these studies, apes received treats when they used the right labels, which makes it hard to tell if they were really sharing in labeling objects or just learning that if they did sign X when they saw a banana, they got a treat, a task any model-free reinforcement learning machine could perform. Extensive analysis of the phrases uttered by these language-able apes demonstrate a low diversity in phrases, meaning they tended to use the exact phrases they learned (e.g., Tickle me) instead of combining words into novel phrases (e.g., I want to be tickled). But in response to these challenges, many point to the Savage-Rumbaugh studies and Kanzi’s incredibly accurate grammatical understanding of commands and playful phrases. The debate is not yet settled.
总体而言，大多数科学家似乎得出这样的结论：一些非人类猿类确实能够学习至少是基本形式的语言，但非人类猿类在这方面的表现比人类差得多，而且不会学习无需刻意训练，这些猿类的能力就无法超越人类幼儿。
On balance, most scientists seem to conclude that some nonhuman apes are indeed capable of learning at least a rudimentary form of language but that nonhuman apes are much worse than humans at it and don’t learn it without painstaking deliberate training. These apes never surpass the abilities of a young human child.
因此，语言似乎在两个方面是人类所独有的。首先，我们天生就倾向于构建和使用语言，而其他动物则没有。其次，我们的语言能力远远超过其他任何动物，即使其他猿类也有可能具备一些基本的符号和语法。
So, language seems unique to humans on two counts. First, we have a natural tendency to construct it and use it, which other animals do not. Second, we have a capacity for language that far surpasses that of any other animal, even if some basic semblance of symbols and grammar is possible in other apes.
但是，如果语言将我们与动物王国的其他成员区分开来，那么，这种看似无害的技巧到底是什么让智人能够登上食物链的顶端？语言的哪些特性让那些使用语言的人如此强大？
But if language is what separates us from the rest of the animal kingdom, then what is it about this seemingly innocuous trick that enabled Homo sapiens to ascend to the top of food chain; what is it about language that makes those who wield it so powerful?
传递思想
Transferring Thoughts
我们独特的语言，带有声明性标签和语法，使大脑群体能够以前所未有的细节和灵活性相互传递内部模拟。人们可以说“从顶部砸碎岩石”或“乔对尤瑟夫很粗鲁”或“记得我们昨天看到的那只狗”，在所有这些情况下，说话者都在刻意选择一个内部模拟的图像和动作来传递给附近的听众。一组n 个大脑只需一个几乎没有任何声音或手势。
Our unique language, with declarative labels and grammar, enables groups of brains to transfer their inner simulations to each other with an unprecedented degree of detail and flexibility. One can say “Smash the rock from the top” or “Joe was rude to Yousef” or “Remember that dog we saw yesterday,” and in all these cases, the talker is deliberately selecting an inner simulation of images and actions to be transferred to nearby listeners. A group of n brains can all re-render the same mental movie of the dog they saw yesterday with merely a few noises or gestures.
当我们谈论这些内在模拟时，特别是在人类的语境中，我们倾向于用概念、想法、思想这样的词语来赋予它们。但所有这些东西都不过是哺乳动物大脑皮层模拟中的渲染。当你“思考”过去或未来的事件时，当你琢磨一只鸟的“概念”时，当你对如何制造一种新工具有了“想法”时，你只是在探索你的大脑皮层构建的丰富的三维模拟世界。从原则上讲，这与老鼠在迷宫中考虑转向哪个方向没有什么不同。概念、想法和思想，就像情景记忆和计划一样，并不是人类独有的。独特之处在于我们能够有意识地将这些内在模拟相互转移，这一技巧只有通过语言才有可能实现。
When we talk about these inner simulations, especially in the context of humans, we tend to imbue them with words like concepts, ideas, thoughts. But all these things are nothing more than renderings in the mammalian neocortical simulation. When you “think” about a past or future event, when you ponder the “concept” of a bird, when you have an “idea” as to how to make a new tool, you are merely exploring the rich three-dimensional simulated world constructed by your neocortex. It is no different, in principle, than a mouse considering which direction to turn in a maze. Concepts, ideas, and thoughts, just like episodic memories and plans, are not unique to humans. What is unique is our ability to deliberately transfer these inner simulations to each other, a trick possible only because of language.
当一只长尾猴发出尖叫声时，附近的所有猴子会迅速从树上跳下来躲起来。显然，这代表了从最先看到鹰的猴子向其他猴子传递信息。但这类传递不详细且不灵活，只能通过遗传上硬编码的信号传递信息。这些信号总是很少，无法根据新情况进行调整或改变。相比之下，语言使说话者能够传递一套极其广泛的内心想法。
When a vervet monkey makes an Eagle nearby! squeal, all nearby monkeys will quickly jump from the trees to hide. Clearly, this represents a transfer of information from the monkey who first saw the eagle to the others. But these kinds of transfers are undetailed and inflexible, capable of transferring information only with genetically hard-coded signals. These signals are always few in number and cannot be adjusted or changed to new situations. In contrast, language enables the talker to transfer an incredibly broad set of inner thoughts.
这种思维转移技巧为早期人类带来了许多实际好处。它使人们能够更准确地教授工具的使用、狩猎技巧和觅食技巧。它使人们能够灵活地协调个体之间的拾荒和狩猎行为——人类可以说，“跟我来，东边两英里处有一具羚羊尸体”或“等在这里，当你听到我吹三声口哨时，我们就可以伏击羚羊了。”
This trick of thought transfer would have provided many practical benefits to early humans. It would have enabled more accurate teaching of tool use, hunting techniques, and foraging tricks. It would have enabled flexible coordination of scavenging and hunting behaviors across individuals—a human could say, “Follow me, there is an antelope carcass two miles east” or “Wait here, let’s ambush the antelope when you hear me whistle three times.”
所有这些实际好处都源于这样一个事实：语言扩大了大脑可以从中提取知识的来源范围。强化的突破使早期脊椎动物能够从自己的实际行动中学习（反复试验）。模拟的突破使早期哺乳动物能够从自己想象的行为中学习（间接反复试验）。心理化的突破使早期灵长类动物能够学习从其他人的实际行为中学习（模仿学习）。但是，语言的突破使得早期人类能够从其他人想象的行为中学习。
All these practical benefits emerge from the fact that language expands the scope of sources a brain can extract learnings from. The breakthrough of reinforcing enabled early vertebrates to learn from their own actual actions (trial and error). The breakthrough of simulating enabled early mammals to learn from their own imagined actions (vicarious trial and error). The breakthrough of mentalizing enabled early primates to learn from other people’s actual actions (imitation learning). But the breakthrough of speaking uniquely enabled early humans to learn from other people’s imagined actions.
学习资源日趋复杂
The Evolution of Progressively More Complex Sources of Learning
|
早期两侧对称动物的强化 REINFORCING IN EARLY BILATERIANS |
早期脊椎动物的模拟 SIMULATING IN EARLY VERTEBRATES |
早期灵长类动物的心智化 MENTALIZING IN EARLY PRIMATES |
早期人类的语言 SPEAKING IN EARLY HUMANS |
学习来源 SOURCE OF LEARNING |
从自己的实际行动中学习 Learning from your own actual actions |
从自己想象的行为中学习 Learning from your own imagined actions |
学习他人的实际行动 Learning from others’ actual actions |
从他人想象的行为中学习 Learning from others’ imagined actions |
向谁学习？ WHO LEARNING FROM? |
你自己 Yourself |
你自己 Yourself |
其他的 Others |
其他的 Others |
行动学习？ ACTION LEARNING FROM? |
实际行动 Actual actions |
想象中的行动 Imagined actions |
实际行动 Actual actions |
想象中的行动 Imagined actions |
语言使我们能够窥视他人的想象，并从中学习——从他们的情景记忆、他们内部模拟的未来行动、他们的反事实。当一个人协调狩猎并说“如果我们集体朝这个方向走，我们就会发现一只羚羊”或“如果我们都等待并伏击，我们就会赢得与野猪的战斗”时，人类正在分享他们自己内心的间接试错的结果，以便整个群体可以从他们的想象中学习。一个拥有山那边狮子情景记忆的人可以通过语言将这种情景记忆传递给其他人。
Language enables us to peer into and learn from the imagination of other minds—from their episodic memories, their internal simulated future actions, their counterfactuals. When a human coordinates a hunt and says, “If we go in this direction as a group we will find an antelope” or “If we all wait and ambush we will win the battle with the boar,” humans are sharing the outcomes of their own inner vicarious trial and errors so that the whole group can learn from their imaginations. One person with an episodic memory of a lion on the other side of a mountain can transfer that episodic memory to others with language.
通过分享我们在想象中所看到的东西，也有可能形成共同的神话，完全虚构的虚构实体和故事仅仅因为它们在我们的大脑中跳跃而流传下来。我们倾向于认为神话是奇幻小说和儿童读物的领域，但它们是现代人类文明的基础。金钱、神、公司和国家都是虚构的概念，只存在于人类大脑的集体想象中。这一想法的早期版本之一是由哲学家约翰·塞尔阐述的，但因尤瓦尔·赫拉利的《人类简史》一书而广为人知。两人认为人类是独一无二的，因为我们“以极其灵活的方式与无数陌生人合作”。在塞尔和赫拉利看来，我们之所以能做到这一点，是因为我们拥有这样的“共同的神话”。用赫拉利的话说：“两个从未谋面的天主教徒可以一起参加十字军东征或集资建造医院，因为他们都信仰上帝”；“两个从未谋面的塞尔维亚人可能会冒着生命危险去拯救对方，因为他们都相信塞尔维亚民族的存在”；“两个从未谋面的律师可以联手为一个完全陌生的人辩护，因为他们都相信法律、正义、人权和正义的存在”。支付的费用。”
By sharing what we see in our imaginations, it is also possible for common myths to form and for entirely made-up imaginary entities and stories to persist merely because they hop between our brains. We tend to think about myths as the province of fantasy novels and children’s books, but they are the foundation of modern human civilizations. Money, gods, corporations, and states are imaginary concepts that exist only in the collective imaginations of human brains. One of the earlier versions of this idea was articulated by the philosopher John Searle, but was famously popularized by Yuval Harari’s book Sapiens. The two argue that humans are unique because we “cooperate in extremely flexible ways with countless numbers of strangers.” And to Searle and Harari, we can do this because we have such “common myths.” In Harari’s words: “Two Catholics who have never met can nevertheless go together on crusade or pool funds to build a hospital because they both believe [in] God” and “Two Serbs who have never met might risk their lives to save one another because both believe in the existence of the Serbian nation” and “Two lawyers who have never met can nevertheless combine efforts to defend a complete stranger because they both believe in the existence of laws, justice, human rights, and money paid out in fees.”
因此，凭借构建共同神话的能力，我们可以协调大量陌生人的行为。这比灵长类动物心理化提供的社会凝聚力系统有了巨大的改进。仅使用心理化来协调行为只有效群体中的每个成员都直接认识彼此。这种合作机制无法扩展；仅靠直接关系维持的人类群体规模的极限估计为大约一百五十人。相比之下，国家、金钱、公司和政府等事物的常见神话使我们能够与数十亿陌生人合作。
And so, with the ability to construct common myths, we can coordinate the behavior of an incredibly large number of strangers. This was a massive improvement over the system of social cohesion provided by primate mentalizing. Coordinating behavior using mentalizing alone works only by each member of a group directly knowing each other. This mechanism of cooperation doesn’t scale; the limit of human group size maintained only by direct relationships has been estimated to be about one hundred fifty people. In contrast, common myths of things like countries, money, corporations, and governments allow us to cooperate with billions of strangers.
Figure 19.1: Maximum number of cooperating individuals based on different cooperation strategies
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
虽然以上所有语言的好处都是事实，但都忽略了更重要的一点。语言的真正天赋并不是优越的教学、合作狩猎或哈拉里的常见神话。这些都不是人类统治世界的原因。如果这些是语言提供的唯一礼物，我们仍然会是狩猎采集猿类，围着篝火跳舞，向水神祈求雨水——当然是顶级捕食者，但绝不是宇航员。语言的这些特征是语言天赋的结果，而不是天赋本身。
While true, all these aforementioned benefits of language miss the larger point. It is not superior teaching, cooperative hunting, or Harari’s common myths that is the true gift of language. None of these are why humans rule the world. If these were the only gifts language offered, we would still be hunter-gatherer apes dancing around campfires praying for rain from the water gods—apex predators, sure, but hardly astronauts. These features of language are consequences of the gift of language, not the gift itself.
类比 DNA 是有用的。DNA 的真正力量不在于它构建的产品（心脏、肝脏、大脑），而在于它促成的过程（进化）。在同样，语言的力量不在于其产品（更好的教学、协调和常见的神话），而在于思想在代代相传、积累和修改的过程。正如基因通过从母细胞跳跃到子细胞而得以存续，思想也通过从一个大脑跳跃到另一个大脑、从一代跳跃到下一代而得以存续。与基因一样，这种跳跃并不统一，而是按照其自身的准进化规则运作——不断选择好的想法并剔除坏的想法。帮助人类生存下来的思想得以存续，而那些没有帮助的思想则消亡了。
An analogy to DNA is useful. The true power of DNA is not the products it constructs (hearts, livers, brains) but the process it enables (evolution). In this same way, the power of language is not its products (better teaching, coordinating, and common myths) but the process of ideas being transferred, accumulated, and modified across generations. Just as genes persist by hopping from parent cell to offspring cell, ideas persist by hopping from brain to brain, from generation to generation. And as with genes, this hopping is not uniform but operates under its own quasi-evolutionary rules—there is a continual selecting of good ideas and pruning of bad ideas. Ideas that helped humans survive persisted, while those that did not perished.
这种思想进化的类比是由理查德·道金斯在他的名著《自私的基因》中提出的。他将这些跳跃的思想称为模因。这个词后来被用来形容在推特上流传的猫咪图片和婴儿照片，但他最初的意思是指在某种文化中人与人之间传播的思想或行为。
This analogy of ideas evolving was proposed by Richard Dawkins in his famous book The Selfish Gene. He called these hopping ideas memes. This word was later appropriated for cat images and baby photos flying around Twitter, but he originally meant them to refer to an idea or behavior that spread from person to person in a culture.
当今人类大脑中存在的知识和行为之所以如此丰富和复杂，是因为其背后的思想在数千年的时间里在无数大脑中积累和修改。甚至数百万代。
The rich complexity of knowledge and behaviors that exist in human brains today are possible only because the underlying ideas have been accumulated and modified across countless brains for thousands, even millions, of generations.
以古代缝制服装的发明为例，人类将死去的动物的皮制成衣服来保暖，许多人认为这项发明早在就像十万年前一样。这项发明之所以能够实现，是因为有无数先前的发明：从动物尸体上切下皮、烘干兽皮、制造绳子和制造骨针。而这些发明本身之所以能够实现，是因为先前发明了锋利的石器。不可能在一瞬间就发明缝制衣服。即使是托马斯·爱迪生也不会这么聪明。爱迪生的新发明是在他获得正确的基础模块之后才出现的。凭借前几代人对电和发电机的理解，他发明了灯泡。
Consider the ancient invention of sewn clothing, in which humans converted hides of dead animals into clothing to keep themselves warm, an invention that many believe emerged as early as one hundred thousand years ago. This invention was possible only because of numerous prior inventions: slicing skin off carcasses, drying hides, manufacturing string, and creating bone needles. And these inventions themselves were possible only due to the prior invention of sharp stone tools. It would never have been possible to invent sewn clothing in a single eureka moment. Not even Thomas Edison would have been so clever. Edison’s new inventions occurred only after he was given the right building blocks. With the understanding of electricity and generators handed to him from prior generations, he invented the lightbulb.
这种积累不仅适用于技术发明，也适用于文化发明。我们传承社会礼仪、价值观、故事、选拔领导人的机制、惩罚的道德规则以及围绕暴力和宽恕的文化信仰。
This accumulation does not apply only to technological inventions but also to cultural ones. We pass down social etiquette, values, stories, mechanisms for selecting leaders, moral rules of punishment, and cultural beliefs around violence and forgiveness.
人类的所有发明，无论是技术发明还是文化发明，都需要积累一些基本构件，然后某个发明者才能恍然大悟。将已有的想法融合成新的东西，并将这一新发明转移给其他人。如果想法的基线总是在一代或两代之后消失，那么一个物种将永远停留在非积累状态，总是一遍又一遍地重新发明同样的想法。动物王国中的所有其他生物都是如此。即使是通过观察学习运动技能的黑猩猩，也不会在几代人之间积累学习成果。
All human inventions, both technological and cultural, require an accumulation of basic building blocks before a single inventor can go “Aha!,” merge the preexisting ideas into something new, and transfer this new invention to others. If the baseline of ideas always fades after a generation or two, then a species will be forever stuck in a nonaccumulating state, always reinventing the same ideas over and over again. This is how it is for all other creatures in the animal kingdom. Even chimpanzees, who learn motor skills through observation, do not accumulate learnings across generations.
这让我们回到第 17 章中看到的模仿实验。让一个四岁的孩子和一只成年黑猩猩观察实验者打开一个拼图盒来获取食物，在这个过程中，他们做了几个不相关的动作。黑猩猩和人类儿童都是通过观察来学习打开拼图盒的；然而，黑猩猩会跳过不相关的步骤，而人类儿童会执行他们观察到的所有步骤，包括不相关的步骤。人类孩子们都是过度模仿者。
This brings us back to the imitation experiments we saw in chapter 17. Take a four-year-old child and an adult chimpanzee and have them observe an experimenter open a puzzle box to get food, in the process performing several irrelevant actions. Both chimps and human children learn to open the puzzle box through observation; however, chimps will skip the irrelevant steps, but human children will perform all the steps they observed, including the irrelevant ones. Human children are over-imitators.
这种过度模仿其实非常聪明。孩子们会根据他们认为老师知道多少来改变模仿的程度——“这个人显然知道自己在做什么，所以她这样做一定有原因。”孩子越不确定老师为什么做某事，他就越有可能准确地模仿老师。模仿所有步骤。此外，他们并不是盲目模仿他们所看到的一切；只有当老师似乎有意模仿时，孩子们才会模仿奇怪的无关行为。如果这个动作看起来是意外，孩子们就会忽略它；孩子们不会模仿老师咳嗽或挠鼻子的动作。如果老师在试图拆开一个新奇的玩具时不断滑倒，孩子会认为这是意外，从而不会模仿这个错误——他们会用更牢固的抓握来成功地将玩具拉开。
This over-imitation is, in fact, quite clever. Children change their degree of copying based on how much they believe the teacher knows—“This person clearly knows what she is doing, so there must be a reason she did that.” The more uncertain a child is about why a teacher is doing something, the more likely he is to exactly copy all the steps. Further, they are not just blindly mirroring whatever they see; children will imitate weird irrelevant behaviors only if their teacher seemed to have intended to do the behavior. If the action appears to be an accident, children will ignore it; children will not copy the coughing or nose scratching of a teacher. If a teacher keeps slipping when trying to pull apart a novel toy, a child will identify this as accidental and thereby not imitate this mistake—they will instead use a firmer grip to successfully pull the toy apart.
虽然这些模仿实验表明人类无需使用语言就可以准确地复制行为，但不可否认的是，语言仍然是我们在复制和传递思想方面的超能力。
While these imitation experiments demonstrate that humans can accurately copy behaviors without using language, it is still undeniably language that is our superpower in the business of copying and transferring ideas.
与专家的沉默模仿相比，使用语言传达如何完成任务提高儿童解决任务的准确性和速度。语言让我们能够压缩信息，使其占用更少的大脑空间，并能更快地从一个大脑转移到另一个大脑。如果我说，“只要你看到一条红蛇，就跑；只要你看到“如果一条绿蛇咬人，就安全了”，这种想法和相应的行为可以立即在整个群体中传播。相比之下，如果每个人都必须通过个人经验或观察别人被多条红蛇咬伤来学习这种“红蛇坏，绿蛇好”的概括，那么将花费更多的时间和脑力。这一事实将不断消失，然后一代又一代地重新学习。没有语言，黑猩猩和其他动物的内在模拟就无法积累，因此，超过给定复杂度阈值的发明——最好的发明——永远无法实现。
Compared to silent imitation of experts, communicating how to do a task using language dramatically improves the accuracy and speed with which children solve tasks. Language lets us condense information so it takes up less brain space and can be more quickly transferred from brain to brain. If I say, “Whenever you see a red snake, run; whenever you see a green snake, it is safe,” that idea and the corresponding behavior can immediately transfer throughout the group. In contrast, if everyone had to learn this “red snake bad, green snake good” generalization through individual experience or through watching someone else get bitten by multiple red snakes, it would take far more time and brainpower. This fact would continually fade and then be relearned from generation to generation. Without language, the inner simulations of chimpanzees and other animals do not accumulate, and thus inventions that are above a given threshold of complexity—the best ones—are forever out of their reach.
奇点已经发生
The Singularity Already Happened
从没有跨代积累到有跨代积累，这是微妙的不连续性改变了一切。在图 19.2中，你可以看到，经过几代人的发展，思想开始变得更加复杂，就像缝制衣服的发明是从更简单的构件组合而成的一样。
Going from no accumulation across generations to some accumulation across generations was the subtle discontinuity that changed everything. In figure 19.2, you can see ideas begin to get more complex across a handful of generations, just as the invention of sewn clothing emerged from a composite of simpler building blocks.
图 19.2
Figure 19.2
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
如果你将时间尺度拉长到数千代人，你就会明白为什么即使只是一些积累也会引发思想复杂性的爆发（如图 19.3所示）。在一段看似永恒停滞的时期，你将在几十万年内看到复杂思想的爆发。
And if you zoom out to the timescale of thousands of generations, you see why even just some accumulation triggers an explosion of idea complexity (as seen in figure 19.3). From a period of seemingly perpetual stasis, you will, in a matter of a few hundred thousand years, get an explosion of complex ideas.
图 19.3
Figure 19.3
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
最终，当所有积累下来的思想总量不再能容纳进单个人的大脑时，思想的复杂性就达到了一个临界点。这导致了无法有效地跨代复制思想。随之而来的有四件事发生，进一步扩大了可以跨代传递的知识范围。首先，人类进化出了更大的大脑，这增加了可以通过个体大脑传承的知识量。其次，人类在群体中变得更加专业化，思想分散在不同成员之间——一些人制造长矛，其他人制造衣服，其他人狩猎，其他人采集食物。第三，人口规模扩大，提供了更多的大脑来存储跨代思想。第四，也是最近也是最重要的，我们发明了书写。书写使人类拥有了思想的集体记忆，可以随意下载，并且有效地包含了无限的知识库。
Eventually, the corpus of ideas accumulated reached a tipping point of complexity when the total sum of accumulated ideas no longer fit into the brain of a single human. This created a problem in sufficiently copying ideas across generations. In response, four things happened that further expanded the extent of knowledge that could be transferred across generations. First, humans evolved bigger brains, which increased the amount of knowledge that can be passed down through individual brains. Second, humans became more specialized within their groups, with ideas distributed across different members—some were the spear makers, others clothing makers, others hunters, others foragers. Third, population sizes expanded, which offered more brains to store ideas across generations. And fourth, most recent and most important, we invented writing. Writing allows humans to have a collective memory of ideas that can be downloaded at will and that can contain effectively an infinite corpus of knowledge.
如果群体没有文字，这种分布式知识就会对群体规模敏感；如果群体规模缩小，大脑不再足够容纳所有信息，知识就会丢失。有证据表明塔斯马尼亚社会就发生过这种情况。八千年前的考古证据表明，塔斯马尼亚人类拥有制作骨制工具、渔网、鱼矛、回旋镖和防寒服的复杂知识。所有这些知识在 19 世纪都丢失了。这种损失似乎起源于海平面上升，塔斯马尼亚人与澳大利亚其他地区的人类群体隔绝，从而大大降低了社会交往人群的人口规模。对于没有文字的人来说，人口越少，知识就越少可以代代相传。
If groups don’t have writing, such distributed knowledge is sensitive to group size; if groups shrink, and there are no longer enough brains to fit all the information into, knowledge is lost. There is evidence that this occurred in societies in Tasmania. Archaeological evidence from eight thousand years ago shows that humans in Tasmania had complex knowledge of making bone tools, nets, fishing spears, boomerangs, and cold-weather clothing. All this knowledge was lost by the 1800s. This loss seems to have been initiated when rising oceans cut the group of humans in Tasmania off from other groups in the rest of Australia, effectively lowering the population size of the group of socially interacting humans. For people without writing, the smaller the population, the less knowledge can persist across generations.
人类之所以独一无二的真正原因是，我们代代相传，积累了共同的模拟（想法、知识、概念、思想）。我们是群脑猿。我们同步我们的内部模拟，将人类文化变成一种元生命形式，其意识在几代人中流经数百万人类大脑的持久想法和思想中得到实例化。这个群脑的基石是我们的语言。
The real reason why humans are unique is that we accumulate our shared simulations (ideas, knowledge, concepts, thoughts) across generations. We are the hive-brain apes. We synchronize our inner simulations, turning human cultures into a kind of meta-life-form whose consciousness is instantiated within the persistent ideas and thoughts flowing through millions of human brains over generations. The bedrock of this hive brain is our language.
语言的出现标志着人类历史的一个转折点，标志着这种新奇独特的进化开始的时间界限：思想的进化。因此，语言的出现与第一个自我复制的 DNA 分子的出现一样具有里程碑意义。语言将人类大脑从一个短暂的器官转变为一个积累发明的永恒媒介。
The emergence of language marked an inflection point in humanity’s history, the temporal boundary when this new and unique kind of evolution began: the evolution of ideas. In this way, the emergence of language was as monumental an event as the emergence of the first self-replicating DNA molecules. Language transformed the human brain from an ephemeral organ to an eternal medium of accumulating inventions.
这些发明包括新技术、新法律、新社会礼仪、新思维方式、新协调系统、新领导人选拔方式、新的暴力与宽恕界限、新价值观、新的共同虚构。语言产生的神经机制早在人们进行数学运算、使用计算机或讨论资本主义的优点之前就出现了。但一旦人类掌握了语言，这些发展几乎是不可避免的。这只是时间问题。事实上，过去几千年来人类的惊人进步与更好的基因无关，而与更好、更复杂的想法的积累有关。
These inventions included new technologies, new laws, new social etiquettes, new ways of thinking, new systems of coordination, new ways of selecting leaders, new thresholds for violence versus forgiveness, new values, new shared fictions. The neurological mechanisms that enable language came far before anyone was doing math, using computers, or discussing the merits of capitalism. But once humans were armed with language, these developments were all but inevitable. It was just a matter of time. Indeed, the incredible ascent of humankind during the past few thousand years had nothing to do with better genes and everything to do with the accumulation of better and more sophisticated ideas.
20
20
大脑中的语言
Language in the Brain
1830 年，一位名叫 Louis Victor Leborgne 的30 岁法国人失去了说话能力。Leborgne 除了会说tan这个音节之外，再也无法说出其他任何声音。Leborgne 的情况比较特殊，因为他的智力基本正常。很明显，当他说话时，他会试图表达某些想法 — — 他会使用手势，改变语气和强调 — — 但唯一发出的声音就是tan。Leborgne 可以理解语言，但他就是无法表达出来。住院多年后，医院里的人都叫他 Tan。
IN 1830, A THIRTY-YEAR-OLD Frenchman by the name of Louis Victor Leborgne lost the ability to speak. Leborgne could no longer say anything other than the syllable tan. What was peculiar about Leborgne’s case was that he was, for the most part, otherwise intellectually typical. It was clear that when he spoke, he was trying to express certain ideas—he would use gestures and alter the tone and emphasis of his speech—but the only sound that ever came out was tan. Leborgne could understand language; he just couldn’t produce it. After many years of hospitalization, he became known around the hospital as Tan.
患者谭去世 20 年后，一位名叫保罗·布罗卡 (Paul Broca) 的法国医生对他的大脑进行了检查，布罗卡对语言神经学特别感兴趣。布罗卡发现莱博涅的大脑左额叶特定且孤立的区域受损。
Twenty years after patient Tan passed away, his brain was examined by a French physician named Paul Broca who had a particular interest in the neurology of language. Broca found that Leborgne had brain damage to a specific and isolated region in the left frontal lobe.
布罗卡有一种预感，认为大脑中有专门负责语言的区域。勒博涅的大脑是布罗卡的第一个线索，证明这个想法可能是正确的。在接下来的两年里，布罗卡煞费苦心地寻找任何最近去世的病人的大脑，这些病人的语言表达能力受损，但保留了其他智力。1865 年，在对 12 个不同的大脑进行尸检后，他发表了现在著名的论文《第三左额叶培养中的语言定位》。结果发现，所有这些病人的大脑皮层左侧都有类似的区域受损，这个区域后来被称为布罗卡区。在过去的一百五十年里，这种现象已经被观察到无数次——如果布罗卡区是受损后，人类就会失去说话的能力，这种情况被称为布罗卡失语症。
Broca had a hunch that there were specific areas in the brain for language. Leborgne’s brain was Broca’s first clue that this idea might be right. Over the next two years, Broca painstakingly sought out the brains of any recently deceased patients who had had impairment in their ability to articulate language but retained their other intellectual faculties. In 1865, after performing autopsies on twelve different brains, he published his now famous paper “Localization of Speech in the Third Left Frontal Cultivation.” It turned out that all of these patients had damage to similar regions on the left side of the neocortex, a region that has come to be called Broca’s area. This has been observed countless times over the past hundred and fifty years—if Broca’s area is damaged, humans lose the ability to produce speech, a condition now called Broca’s aphasia.
布罗卡完成这项工作几年后，德国医生卡尔·韦尼克 (Carl Wernicke) 被另一组语言障碍所困扰。韦尼克发现，与布罗卡不同，患者虽然说话正常，但缺乏表达能力理解言语。这些患者会说出完整的句子，但句子毫无意义。例如，这样的患者可能会说“你知道 smoodle pinker 病了，我想带他过来照顾他就像你以前想要的那样。”
Several years after Broca did his work, Carl Wernicke, a German physician, was perplexed by a different set of language difficulties. Wernicke found patients who, unlike Broca’s, could speak fine but lacked the ability to understand speech. These patients would produce whole sentences, but the sentences made no sense. For example, such a patient might say something like “You know that smoodle pinkered and that I want to get him round and take care of him like you want before.”
韦尼克遵循布罗卡的策略，在这些患者的大脑中也发现了一个受损区域。这个区域也位于左侧，但位于大脑皮层后部更靠后的位置，现在这个区域被称为韦尼克区。韦尼克区受损会导致韦尼克失语症，这是一种患者失去理解语言能力的疾病。
Wernicke, following Broca’s strategy, also found a damaged area in the brains of these patients. It was also on the left side but farther back in the posterior neocortex, a region now dubbed Wernicke’s area. Damage to Wernicke’s causes Wernicke’s aphasia, a condition in which patients lose the ability to understand speech.
布罗卡区和韦尼克区的一个显著特点是，它们的语言功能并不只针对特定的语言模式，而是选择性语言。患有布罗卡氏失语症的患者说话时会受到同样多的障碍，就像他们说话时一样。无法用书写文字。主要使用手语交流的患者在以下情况下会失去流利地使用手语的能力：布罗卡区受损。韦尼克区受损会导致理解能力下降口语和书面语。事实上，当听力健全的人听别人说话时，这些相同的语言区域就会被激活，而当聋哑人观察某人的手语。布罗卡区不选择性地表达、书写或手语；它选择性地产生语言的一般能力。而韦尼克区不选择性地聆听、阅读或观察手语；它选择性地理解语言的一般能力。
A revealing feature of both Broca’s and Wernicke’s areas is that their language functions are not selective for only certain modalities of language, but rather are selective for language in general. Patients with Broca’s aphasia become equally impaired in speaking words as they are in writing words. Patients who primarily communicate using sign language lose their ability to sign fluently when Broca’s area is damaged. Damage to Wernicke’s area produces deficits in understanding both spoken language and written language. Indeed, these same language areas are activated when a hearing-abled person listens to someone speak and when a deaf person watches someone sign. Broca’s area is not selective for verbalizing, writing, or signing; it is selective for the general ability to produce language. And Wernicke’s area is not selective for listening, reading, or watching signs; it is selective for the general ability to understand language.
图 20.1
Figure 20.1
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
人类运动皮层与脑干区域直接连接，控制喉部和声带——这是人类大脑与其他猿类大脑之间为数不多的结构差异之一。人类大脑皮层可以独特地控制声带，这无疑是使用口头语言的一种适应。但这只是试图理解语言进化的借口；这种独特的回路并不是语言进化的突破。我们之所以知道这一点，是因为人类学习非语言时，可以像学习口头语言一样流利和轻松——语言并不是需要这种声带连接的技巧。人类对喉部的独特控制要么与语言的其他变化共同进化，要么在语言变化之后进化（从手势语言过渡到口头语言），要么在语言变化之前进化（适应其他非语言目的）。无论如何，人类对喉部的控制并不是语言进化的根本原因。
The human motor cortex has a unique connection directly to the brainstem area for controlling the larynx and vocal cords—this is one of the few structural differences between the brains of humans and those of other apes. The human neocortex can uniquely control the vocal cords, which is surely an adaptation for using verbal language. But this is a red herring in trying to understand the evolution of language; this unique circuitry is not the evolutionary breakthrough that enabled language. We know this because humans can learn nonverbal language with as much fluency and ease as they learn verbal language—language is not a trick that requires this wiring with the vocal cords. Humans’ unique control of the larynx either coevolved with other changes for language in general, evolved after them (to transition from a gesture-like language to a verbal language), or evolved before them (adapted for some other nonlanguage purpose). In any case, it is not human control of the larynx that enabled language.
布罗卡和韦尼克的发现表明，语言源自大脑中的特定区域，并且包含在几乎总是位于大脑皮层左侧的子网络中。语言的特定区域也有助于解释语言能力可能与其他智力能力完全脱节的事实。许多语言障碍的人智力上正常。有些人在语言上很有天赋，但在智力上却有缺陷。1995 年，两位研究人员 Neil Smith 和 Ianthi-Maria Tsimpli 发表了他们对一位名叫 Christopher 的儿童语言天才的研究。Christopher 的认知能力严重受损，手眼协调能力很差，连扣衬衫纽扣等基本任务都做不完，也无法弄清楚如何赢得井字游戏或跳棋游戏。但在语言方面，Christopher 是超人：他能读、能写、能说超过十五种语言。尽管他的大脑其余部分“受损”，但他的语言区域不仅没有受到影响，而且非常出色。关键在于语言不是从整个大脑中产生的，而是从特定的子系统中产生的。
Broca’s and Wernicke’s discoveries demonstrated that language emerges from specific regions in the brain and that it is contained in a subnetwork almost always found on the left side of the neocortex. The specific regions for language also help explain the fact that language capacity can be quite dissociated from other intellectual capacities. Many people who become linguistically impaired are otherwise intellectually typical. And people can be linguistically gifted while otherwise intellectually impaired. In 1995, two researchers, Neil Smith and Ianthi-Maria Tsimpli, published their research on a child language savant named Christopher. Christopher was extremely cognitively impaired, had terrible hand-eye coordination, would struggle to do basic tasks such as buttoning a shirt, and was incapable of figuring out how to win a game of tic-tac-toe or checkers. But Christopher was superhuman when it came to language: he could read, write, and speak over fifteen languages. Although the rest of his brain was “impaired,” his language areas were not only spared but were brilliant. The point is that language emerges not from the brain as a whole but from specific subsystems.
这表明语言并不是拥有更多大脑皮层的必然结果。语言并不是人类通过扩大黑猩猩的大脑而“免费”获得的东西。语言是一种特定且独立的技能，是进化过程中融入我们大脑的。
This suggests that language is not an inevitable consequence of having more neocortex. It is not something humans got “for free” by virtue of scaling up a chimpanzee brain. Language is a specific and independent skill that evolution wove into our brains.
所以这似乎可以结案了。我们已经发现了人类大脑的语言器官：人类进化出了大脑皮层的两个新区域——布罗卡区和韦尼克区——它们连接在一起形成一个专门用于语言的特定子网络。这个子网络赋予了我们语言，这就是为什么人类有语言而其他猿类没有的原因。结案了。
So this would seem to close the case. We have found the language organ of the human brain: humans evolved two new areas of neocortex—Broca’s and Wernicke’s areas—which are wired together into a specific subnetwork specialized for language. This subnetwork gifted us language, and that is why humans have language and other apes don’t. Case closed.
不幸的是，故事并不是那么简单。
Unfortunately, the story is not so simple.
笑声还是语言？
Laughs or Language?
以下事实使事情变得复杂：你的大脑和黑猩猩的大脑实际上是完全相同的；人类的大脑几乎完全一样，只是放大的黑猩猩大脑。这包括被称为布罗卡区和韦尼克区的区域。这些区域不是在早期人类中进化而来的；它们在早期灵长类动物中出现得更早。它们是随着心理化的突破而出现的大脑皮层区域的一部分。黑猩猩、倭黑猩猩，甚至猕猴都拥有这些区域，并且具有几乎相同的连接性。因此，布罗卡区或韦尼克区的出现并不是人类语言天赋的来源。
The following fact complicates things: Your brain and a chimpanzee brain are practically identical; a human brain is, almost exactly, just a scaled-up chimpanzee brain. This includes the regions known as Broca’s area and Wernicke’s area. These areas did not evolve in early humans; they emerged much earlier, in the first primates. They are part of the areas of the neocortex that emerged with the breakthrough of mentalizing. Chimpanzees, bonobos, and even macaque monkeys all have exactly these areas with practically identical connectivity. Thus, it was not the emergence of Broca’s or Wernicke’s areas that gave humans the gift of language.
也许人类语言是对现有猿类交流系统的进一步发展？这或许可以解释为什么其他灵长类动物的大脑中仍然存在这些语言区域。黑猩猩、倭黑猩猩和大猩猩都有复杂的手势和叫声来表示不同的意思。翅膀是从手臂进化而来的，多细胞生物是从单细胞生物进化而来的，所以如果人类语言是从我们猿类祖先更原始的交流系统进化而来的，那就说得通了。但这并不是语言在大脑中进化的方式。
Perhaps human language was an elaboration on the existing system of ape communication? This might explain why these language areas are still present in other primates. Chimpanzees, bonobos, and gorillas all have sophisticated suites of gestures and hoots that signal different things. Wings evolved from arms, and multicellular organisms evolved from single-celled organisms, so it would make sense if human language evolved from the more primitive communication systems of our ape ancestors. But this is not how language evolved in the brain.
在其他灵长类动物中，大脑皮层的这些语言区域是存在的，但与交流无关。如果你损坏猴子的布罗卡区和韦尼克区，它就没有对猴子交流的影响。如果人类的这些基因被破坏，我们就会完全失去使用语言的能力。
In other primates, these language areas of the neocortex are present but have nothing to do with communication. If you damage Broca’s and Wernicke’s areas in a monkey, it has no impact on monkey communication. If you damage them in humans, we lose the ability to use language entirely.
当我们将猿类手势与人类语言进行比较时，我们其实是在比较苹果和橘子。它们在交流方面的共同用途掩盖了一个事实：它们是完全不同的神经系统，彼此之间没有任何进化关系。
When we compare ape gestures to human language, we are comparing apples to oranges. Their common use for communication obscures the fact that they are entirely different neurological systems without any evolutionary relationship to each other.
事实上，人类继承了与猿类完全相同的交流系统，但那不是我们的语言，而是我们的情感表达。
Humans have, in fact, inherited the exact same communication system of apes, but it isn’t our language—it is our emotional expressions.
20 世纪 90 年代中期，一位 50 多岁的老师发现自己说话困难。三天后，他的症状恶化。等他去看医生时，他的右脸已经瘫痪，说话变得缓慢、含糊不清。当他被要求微笑时，他只有一侧脸会动，导致他露出不对称的笑容（图 20.2）。
In the mid-1990s, a teacher in his fifties noticed that he was struggling to speak. Over the course of three days, his symptoms worsened. By the time he made it to the doctor, the right side of his face was paralyzed, and the man’s speech was slowed and slurred. When he was asked to smile, only one side of his face would move, leading to a lopsided smirk (figure 20.2).
在检查这名男子时，医生发现了一些令人困惑的事情。当医生讲笑话或说一些真正令人愉快的话时，这名男子可以很好地微笑。当他笑的时候，他的左脸可以正常运作，但当他被要求自发微笑时，他却做不到。
When examining the man, the doctor noticed something perplexing. When the doctor told a joke or said something genuinely pleasant, the man could smile just fine. The left side of his face worked normally when he was laughing, but when he was asked to smile voluntarily, the man was unable to do it.
人类大脑对面部表情有平行的控制；有一个较古老的情绪表达系统，其中有情绪状态和反射反应之间的硬编码映射。这个系统由杏仁核等古老结构控制。还有一个单独的系统提供对面部肌肉的自愿控制，该系统由大脑皮层控制。
The human brain has parallel control of facial expressions; there is an older emotional-expression system that has a hard-coded mapping between emotional states and reflexive responses. This system is controlled by ancient structures like the amygdala. Then there is a separate system that provides voluntary control of facial muscles that is controlled by the neocortex.
图 20.2：一名患者的运动皮质与左侧面部之间的连接受损，但杏仁核与左侧面部之间的连接完好无损
Figure 20.2: A patient with a damaged connection between motor cortex and the left side of face, but an intact connection between the amygdala and left side of face is spared
图片来自 Trepel 等人，1996 年。已获许可使用。
Images from Trepel et al., 1996. Used with permission.
原来，这位老师的脑干受到损伤，导致大脑皮层与左脸肌肉之间的联系中断，但杏仁核与左脸肌肉之间的联系却没有受到影响。这意味着他无法自主控制左脸，但他的情绪表达系统可以很好地控制他的脸。虽然他无法自主抬起眉毛，但他却能够笑、皱眉和哭。
It turned out that this teacher had a lesion in his brain stem that had disrupted the connection between his neocortex and the muscles on the left side of his face but had spared the connection between his amygdala and those same muscles. This meant that he couldn’t voluntarily control the left side of his face, but his emotional-expression system could control his face just fine. While he was unable to voluntarily lift an eyebrow, he was eminently able to laugh, frown, and cry.
患有严重布罗卡失语症和韦尼克失语症的人也有同样的表现。即使是一个词都说不出来的人也会笑和哭。为什么？因为情绪表达来自一个与语言完全不同的系统。
This is also seen in individuals with severe forms of Broca’s and Wernicke’s aphasia. Even individuals who can’t utter a single word can readily laugh and cry. Why? Because emotional expressions emerge from a system entirely separate from language.
类人猿与人类交流的同类比较是类人猿发声与人类情感表达的比较。简单来说：其他灵长类动物只有一个交流系统，即情感表达系统，位于杏仁核和脑干等较老的区域。它将情绪状态映射到交流手势和声音上。事实上，正如珍·古道尔所注意到的，“在没有适当情绪状态的情况下发出声音似乎几乎是一种（对黑猩猩来说）这是不可能完成的任务。”这种情绪表达系统由来已久，可以追溯到早期哺乳动物，甚至更早。然而，人类有两种交流系统——我们有同样古老的情感表达系统，我们有新进化的语言大脑皮层中的系统。
The apples-to-apples comparison between ape and human communication is between ape vocalizations and human emotional expressions. To simplify a bit: Other primates have a single communication system, their emotional-expression system, located in older areas like the amygdala and brainstem. It maps emotional states to communicative gestures and sounds. Indeed, as noticed by Jane Goodall, “the production of a sound in the absence of the appropriate emotional state seems to be an almost impossible task [for chimpanzees].” This emotional-expression system is ancient, going back to early mammals, perhaps even earlier. Humans, however, have two communication systems—we have this same ancient emotional expression system and we have a newly evolved language system in the neocortex.
图 20.3
Figure 20.3
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
人类的笑声、哭声和皱眉是古老而原始的交流系统的进化遗留，猿猴的叫声和手势就是从这个系统演化而来的。然而，当我们说话时，我们所做的事与猿猴的任何交流系统都没有任何明显的相似之处。
Human laughs, cries, and scowls are evolutionary remnants of an ancient and more primitive system for communication, a system from which ape hoots and gestures emerge. However, when we speak words, we are doing something without any clear analog to any system of ape communication.
这就解释了为什么猴子布罗卡区和韦尼克区受损对交流完全没有影响。猴子仍然可以发出叫声，就像受到这种损伤的人虽然无法说出一个连贯的词，但仍然可以笑、哭、微笑、皱眉和皱眉一样。猴子的手势是自动的情绪表达，并非来自大脑皮层；它们更像是人类的笑声而不是语言。
This explains why lesions to Broca’s and Wernicke’s areas in monkeys have absolutely no impact on communication. A monkey can still hoot and holler for the same reason a human with such damage can still laugh, cry, smile, frown, and scowl even while he can’t utter a single coherent word. The gestures of monkeys are automatic emotional expressions and don’t emerge from the neocortex; they are more like a human laugh than language.
情绪表达系统和语言系统还有另一个区别：一个是遗传上固定的，另一个是后天习得的。人类和其他猿类共有的情绪表达系统在很大程度上是遗传上固定的。有证据表明，在孤立环境中长大的猴子最终还是会产生所有的情绪正常的手势行为，黑猩猩和倭黑猩猩有近 90% 的相同之处相同的手势。同样，世界各地的人类文化和儿童在情绪表达方面也有显著的重叠，这表明我们的情绪表达至少有一部分是基因硬编码的，而不是后天习得的。所有人类（即使是天生失明和失聪的人）都会以相对相似的方式哭泣、微笑、大笑和皱眉来回应达到相似的情绪状态。
The emotional-expression system and the language system have another difference: one is genetically hardwired, and the other is learned. The shared emotional-expression system of humans and other apes is, for the most part, genetically hardwired. As evidence, monkeys who are raised in isolation still end up producing all of their normal gesture-call behavior, and chimpanzees and bonobos share almost 90 percent of the same gestures. Similarly, human cultures and children from around the world have remarkable overlap in emotional expressions, suggesting that at least some parts of our emotional expressions are genetically hard-coded and not learned. All human beings (even those born blind and deaf) cry, smile, laugh, and frown in relatively similar ways in response to similar emotional states.
然而，人类较新的语言系统对学习极其敏感——如果一个孩子长时间没有学习语言，他或她将无法掌握语言在以后的生活中会逐渐形成。与天生的情绪表达不同，语言的特征在不同文化中差异很大。事实上，一个出生时没有任何新皮层的人类婴儿仍然会以通常的方式表达这些情绪但永远不会说话。
However, the newer language system in humans is incredibly sensitive to learning—if a child goes long enough without being taught language, he or she will be unable to acquire it later in life. Unlike innate emotional expressions, features of language differ greatly across cultures. And indeed, a human baby born without any neocortex will still express these emotions in the usual way but will never speak.
这就是语言的神经生物学难题。语言不是从某种新进化的结构中产生的。语言不是从人类独特的新皮层对喉部和面部的控制中产生的。（尽管这确实使更复杂的言语表达成为可能）。语言并非源自早期猿类交流系统的某些改进。然而，语言是全新的。
So here is the neurobiological conundrum of language. Language did not emerge from some newly evolved structure. Language did not emerge from humans’ unique neocortical control over the larynx and face (although this did enable more complex verbalizations). Language did not emerge from some elaboration of the communication systems of early apes. And yet, language is entirely new.
那么解锁什么语言？
So what unlocked language?
语言课程
The Language Curriculum
所有的鸟都知道如何飞翔。这是否意味着所有的鸟都具有遗传上固定的飞行知识？嗯，不是。鸟不是生来就知道如何飞翔；所有的幼鸟都必须独立学习飞行。它们从拍打翅膀开始，试图盘旋，第一次尝试滑翔，最终，经过足够的重复，它们学会了。但如果飞行不是遗传上固定的，那么为什么大约 100% 的幼鸟都能独立学习如此复杂的技能呢？
All birds know how to fly. Does this mean that all birds have genetically hardwired knowledge of flying? Well, no. Birds are not born knowing how to fly; all baby birds must independently learn how to fly. They start by flapping wings, trying to hover, making their first glide attempt, and eventually, after enough repetitions, they figure it out. But if flying is not genetically hard-coded, then how is it that approximately 100 percent of all baby birds independently learn such a complex skill?
像飞行这样复杂的技能信息量过于密集，无法直接硬编码到基因组中。编码通用学习系统（如大脑皮层）和特定的硬连线学习课程（跳跃的本能、拍打翅膀的本能和试图滑翔的本能）更有效。学习系统和课程的结合使每一只幼鸟都能学会如何飞翔。
A skill as sophisticated as flying is too information-dense to hard-code directly into a genome. It is more efficient to encode a generic learning system (such as a cortex) and a specific hardwired learning curriculum (instinct to want to jump, instinct to flap wings, and instinct to attempt to glide). It is the pairing of a learning system and a curriculum that enables every single baby bird to learn how to fly.
在人工智能领域，课程的力量和重要性众所周知。20 世纪 90 年代，加州大学圣地亚哥分校的语言学家兼认知科学教授 Jeffrey Elman 是第一批使用神经网络来预测句子中下一个单词的人之一给定前面的单词。学习策略很简单：不断向神经网络展示一个又一个单词、一个又一个句子，让它根据前面的单词预测下一个单词，然后每次将网络中的所有权重推向正确答案。从理论上讲，它应该能够正确预测它从未见过的新句子中的下一个单词。
In the world of artificial intelligence, the power and importance of curriculum is well known. In the 1990s, a linguist and professor of cognitive sciences at UC San Diego, Jeffrey Elman, was one of the first to use neural networks to try to predict the next word in a sentence given the previous words. The learning strategy was simple: Keep showing the neural network word after word, sentence after sentence, have it predict the next word based on the prior words, then nudge all the weights in the network toward the right answer each time. Theoretically, it should have been able to correctly predict the next word in a novel sentence it had never seen before.
它不起作用。
It didn’t work.
然后，Elman 尝试了不同的方法。他没有同时向神经网络展示所有复杂程度的句子，而是先向它展示极其简单的句子，只有当网络在这些句子上表现良好后，他才会提高复杂程度。换句话说，他设计了一个课程。事实证明，这个课程很有效。经过这个课程的训练后，他的神经网络可以正确地完成复杂的句子。
Then Elman tried something different. Instead of showing the neural network sentences of all levels of complexity at the same time, he first showed it extremely simple sentences, and only after the network performed well at these did he increase the level of complexity. In other words, he designed a curriculum. And this, it turned out, worked. After being trained with this curriculum, his neural network could correctly complete complex sentences.
为人工智能设计课程的想法不仅适用于语言，还适用于多种类型的学习。还记得我们在突破 #2中看到的无模型强化算法 TD-Gammon吗？TD-Gammon 使计算机在西洋双陆棋游戏中胜过人类。我遗漏了 TD-Gammon 训练的一个关键部分。它没有通过与人类专家进行无休止的西洋双陆棋游戏的反复试验来学习。如果它这样做了，它就永远不会学会，因为它永远不会赢。TD-Gammon 是通过与自己对战来训练的。TD-Gammon 总是有一个势均力敌的玩家。这是训练强化学习系统的标准策略。谷歌的 AlphaZero 也是通过与自己对战来训练的。用于训练模型的课程同样重要作为模型本身。
This idea of designing a curriculum for AI applies not just to language but to many types of learning. Remember the model-free reinforcement algorithm TD-Gammon that we saw in breakthrough #2? TD-Gammon enabled a computer to outperform humans in the game of backgammon. I left out a crucial part of how TD-Gammon was trained. It did not learn through the trial and error of endless games of backgammon against a human expert. If it had done this, it would never have learned, because it would never have won. TD-Gammon was trained by playing against itself. TD-Gammon always had an evenly matched player. This is the standard strategy for training reinforcement learning systems. Google’s AlphaZero was also trained by playing itself. The curriculum used to train a model is as crucial as the model itself.
要教授一项新技能，改变课程往往比改变学习系统更容易。事实上，这似乎是进化在实现复杂技能时反复采用的解决方案——猴子攀爬、鸟类飞行，甚至人类语言似乎都是这样运作的。它们都来自新进化的硬连线课程。
To teach a new skill, it is often easier to change the curriculum instead of changing the learning system. Indeed, this is the solution that evolution seems to have repeatedly settled on when enabling complex skills—monkey climbing, bird flying, and, yes, even human language all seem to work this way. They emerge from newly evolved hardwired curriculums.
早在人类婴儿开始用语言进行对话之前，他们就开始进行所谓的原始对话。到四个月大时，在婴儿开口说话之前，他们就会轮流与父母来回发声，面部表情和手势。研究表明，婴儿会与母亲的停顿时间保持一致，从而形成轮流说话的节奏；婴儿会发出声音、停顿、倾听父母并等待父母的回应。看来，交谈并不是学习语言能力的自然结果；相反，学习语言的能力至少在一定程度上是一种更简单的遗传本能，即参与交谈。语言似乎是建立在这种手势和声音轮流说话的固定课程之上的。这种轮流说话的方式首先在早期人类中进化而来；黑猩猩婴儿没有表现出这种行为。
Long before human babies engage in conversations using words, they engage in what are called proto-conversations. By four months of age, long before babies speak, they will take turns with their parents in back-and-forth vocalizations, facial expressions, and gestures. It has been shown that infants will match the pause duration of their mothers, thereby enabling a rhythm of turn-taking; infants will vocalize, pause, attend to their parents, and wait for their parents’ response. It seems conversation is not a natural consequence of the ability to learn language; rather, the ability to learn language is, at least in part, a consequence of a simpler genetically hard-coded instinct to engage in conversation. It seems to be this hardwired curriculum of gestural and vocal turn-taking on which language is built. This type of turn-taking evolved first in early humans; chimpanzee infants show no such behavior.
到九个月大时，在学会说话之前，人类婴儿就开始表现出第二种新奇的行为：共同注意物体。当当母亲看着或指向一个物体时，人类婴儿会将注意力集中在同一个物体上，并使用各种非语言机制来确认她看到了母亲看到的东西。这些非语言确认可以很简单，比如婴儿一边微笑一边在物体和母亲之间来回看，抓住它并把它递给母亲，或者只是指着它并看着回到她母亲身边。
By nine months of age, still before speech, human infants begin to demonstrate a second novel behavior: joint attention to objects. When a mother looks at or points to an object, a human infant will focus on that same object and use various nonverbal mechanisms to confirm that she saw what her mother saw. These nonverbal confirmations can be as simple as the baby looking back and forth between the object and her mother while smiling, grasping it and offering it to her mother, or just pointing to it and looking back at her mother.
科学家们已经竭尽全力证实，这种行为并非为了获得物体或从父母那里得到积极的情绪反应，而是真正试图与他人分享注意力。例如，指向一个物体的婴儿会继续指向它，直到她的父母将目光在同一物体和婴儿之间来回移动。如果父母只是看着婴儿并热情地说话，或者看着物体但没有回头看婴儿（确认她看到了婴儿看到的东西），婴儿就会不满意并再次指出。婴儿经常会因为这种确认而感到满意，而没有得到他们关注的对象，这一事实强烈表明他们的意图不是获得该对象，而是与母亲一起关注。
Scientists have gone to great lengths to confirm that this behavior is not an attempt to obtain the object or get a positive emotional response from their parents, but instead is a genuine attempt to share attention with others. For example, an infant who points to an object will continue pointing to it until her parent alternates their gaze between the same object and the infant. If the parent simply looks at the infant and speaks enthusiastically or looks at the object but doesn’t look back at the infant (confirming she saw what the infant saw), the infant will be unsatisfied and point again. The fact that infants frequently are satisfied by this confirmation without being given the object of their attention strongly suggests their intent was not to obtain the object but to engage in joint attention with their mothers.
就像原始对话一样，这种语言形成前的共同注意行为似乎是人类婴儿所独有的；非人类灵长类动物不会参与共同注意。黑猩猩对确保其他人注意自己的行为没有兴趣共同注意和追随凝视之间有一个关键的区别。许多动物，甚至是海龟，都被证明会追随同类的目光。如果一只海龟朝某个方向看，附近的海龟通常也会这样做。但这只能用看别人看的地方的反射来解释。然而，共同注意是一个更为刻意的过程，它来回确认两个人的思想都在关注同一个外部物体。
Like proto-conversations, this pre-language behavior of joint attention seems to be unique to human infants; nonhuman primates do not engage in joint attention. Chimpanzees show no interest in ensuring someone else attends to the same object they do. They will, of course, follow the gaze of others around them—looking in the direction they see others look. But there is a crucial distinction between joint attention and gaze following. Lots of animals, even turtles, have been shown to follow the gaze of another of their own species. If a turtle looks in a certain direction, nearby turtles will often do the same. But this can be explained merely by a reflex to look where others look. Joint attention, however, is a more deliberate process of going back and forth to confirm that both minds are attending to the same external object.
孩子们天生具有进行原始对话和共同注意的古怪能力有什么意义呢？这不是为了模仿学习；非人类灵长类动物在没有原始对话或共同注意的情况下也能很好地进行模仿学习。这不是为了建立社会纽带；非人类灵长类动物和其他哺乳动物有许多其他建立社会纽带的机制。似乎共同注意和原始对话进化只有一个原因。父母与孩子达到共同关注的状态后，他们做的第一件事是什么？他们给事物贴上标签。
What’s the point of children’s quirky prewired ability to engage in proto-conversations and joint attention? It is not for imitation learning; nonhuman primates engage in imitation learning just fine without proto-conversations or joint attention. It is not for building social bonds; nonhuman primates and other mammals have plenty of other mechanisms for building social bonds. It seems that joint attention and proto-conversations evolved for a single reason. What is one of the first things that parents do once they have achieved a state of joint attention with their child? They assign labels to things.
一岁时婴儿表现出的共同注意力越多，孩子的词汇量是在十二个月后开始的。一旦人类婴儿开始学习单词，他们就会自然地将这些单词组合起来形成语法句子。通过原始对话和共同注意的硬连线系统建立陈述性标签的基础，语法使他们能够将这些单词组合成句子，然后可以构造出完整的故事和想法。
The more joint attention expressed by an infant at the age of one year, the larger the child’s vocabulary is twelve months later. Once human infants begin to learn words, they start naturally combining these words to form grammatical sentences. With the foundation of declarative labels in place through the hardwired systems of proto-conversations and joint attention, grammar allows them to combine these words into sentences, which can then be constructed to create entire stories and ideas.
人类可能也进化出了一种独特的本能，即通过提问来了解他人的内心模拟。即使是 Kanzi、Washoe 和其他获得了令人印象深刻的复杂语言能力的猿类，也从未问过最简单的关于别人的问题。他们会要求食物和玩耍，但不会询问他人的内心世界。甚至在人类儿童能够构造语法句子之前，他们就会问别人问题：“想要这个吗？”“饿了吗？”所有语言都使用相同的升调询问是非问题时。当你听到某人用你听不懂的语言说话时，你仍然可以识别出你正在被问问题。这种理解如何指定问题的本能也可能是关键部分我们的语言课程。
Humans may have also evolved a unique hardwired instinct to ask questions to inquire about the inner simulations of others. Even Kanzi, Washoe, and the other apes that acquired impressively sophisticated language abilities never asked even the simplest questions about others. They would request food and play but would not inquire about another’s inner mental world. Even before human children can construct grammatical sentences, they will ask others questions: “Want this?” “Hungry?” All languages use the same rising intonation when asking yes/no questions. When you hear someone speak in a language you do not understand, you can still identify when you are being asked a question. This instinct to understand how to designate a question may also be a key part of our language curriculum.
因此，我们并没有意识到，当我们开心地和婴儿来回发出语无伦次的咿呀声（原始对话）时，当我们来回传递物品并微笑（共同注意）时，当我们提出并回答婴儿提出的甚至是无意义的问题时，我们正在不知不觉地执行一种进化上硬编码的学习程序，该程序旨在赋予人类婴儿语言天赋。这就是为什么被剥夺与他人接触的人类会发展出情感表达，但永远不会发展出语言的原因。语言课程需要老师和学生。
So we don’t realize it, but when we happily go back and forth making incoherent babbles with babies (proto-conversations), when we pass objects back and forth and smile (joint attention), and when we pose and answer even nonsensical questions from infants, we are unknowingly executing an evolutionarily hard-coded learning program designed to give human infants the gift of language. This is why humans deprived of contact with others will develop emotional expressions, but they’ll never develop language. The language curriculum requires both a teacher and a student.
随着这种本能学习课程的实施，年轻人的大脑会将新皮质中较旧的心理化区域重新用于语言的新用途。布罗卡区或韦尼克区并不是新的，而是将其重新用于语言的底层学习程序。布罗卡区或韦尼克区没有什么特别之处的证据是：整个左半球被切除的儿童仍然可以学习语言能力很强，并且会重新利用大脑右侧新皮质的其他区域来执行语言。事实上，大约 10% 的人，无论出于什么原因，都倾向于使用大脑右侧而不是左侧来执行语言。新的研究甚至质疑布罗卡区和韦尼克区实际上是语言所在地的观点；语言区域可能位于整个新皮质，甚至位于基底神经节中。
And as this instinctual learning curriculum is executed, young human brains repurpose older mentalizing areas of the neocortex for the new purpose of language. It isn’t Broca’s or Wernicke’s areas that are new, it is the underlying learning program that repurposes them for language that is new. As proof that there is nothing special about Broca’s or Wernicke’s areas: Children with the entire left hemisphere removed can still learn language just fine and will repurpose other areas of the neocortex on the right side of the brain to execute language. In fact, about 10 percent of people, for whatever reason, tend to use the right side of the brain, not the left, for language. Newer studies are even calling into question the idea that Broca’s and Wernicke’s areas are actually the loci of language; language areas may be located all over the neocortex and even in the basal ganglia.
重点是：人类大脑中没有语言器官，就像鸟类大脑中没有飞行器官一样。问语言在大脑中的位置可能和问打棒球或弹吉他在大脑中的位置一样愚蠢。这些复杂的技能并不局限于特定区域；它们来自许多区域的复杂相互作用。使这些技能成为可能的不是执行这些技能的单个区域，而是一个迫使复杂区域网络协同学习这些技能的课程。
Here is the point: There is no language organ in the human brain, just as there is no flight organ in the bird brain. Asking where language lives in the brain may be as silly as asking where playing baseball or playing guitar lives in the brain. Such complex skills are not localized to a specific area; they emerge from a complex interplay of many areas. What makes these skills possible is not a single region that executes them but a curriculum that forces a complex network of regions to work together to learn them.
所以这就是为什么你的大脑和黑猩猩的大脑几乎一模一样，但只有人类才有语言。人类大脑的独特之处并不在大脑皮层中；独特之处是隐藏的、微妙的，藏在杏仁核和脑干等较老的结构深处。这是对硬连线本能的调整，这种本能让我们轮流说话，让孩子和父母来回盯着看，让我们提出问题。
So this is why your brain and a chimp brain are practically identical and yet only humans have language. What is unique in the human brain is not in the neocortex; what is unique is hidden and subtle, tucked deep in older structures like the amygdala and brain stem. It is an adjustment to hardwired instincts that makes us take turns, makes children and parents stare back and forth, and that makes us ask questions.
这也是猿类能够学习语言基础知识的原因。猿类大脑皮层在这方面的能力非常出色。猿类很难掌握语言，因为它们缺乏学习语言所需的本能。很难让黑猩猩参与到共同关注中；很难让它们轮流说话；它们也没有分享想法或提问的本能。如果没有这些本能，语言就很难掌握，就像没有跳跃本能的鸟永远学不会飞一样。
This is also why apes can learn the basics of language. The ape neocortex is eminently capable of it. Apes struggle to become sophisticated at it merely because they don’t have the required instincts to learn it. It is hard to get chimps to engage in joint attention; it is hard to get them to take turns; and they have no instinct to share their thoughts or ask questions. And without these instincts, language is largely out of reach, just as a bird without the instinct to jump would never learn to fly.
所以，总结一下：我们知道，人类大脑与众不同的突破是语言。语言之所以强大，是因为它让我们能够从其他人的想象中学习，并让思想在几代人中积累。我们知道，语言是通过一种固定的学习课程在人类大脑中形成的，这种课程将旧的心理新皮层区域重新用于语言区域。
So, to recap: We know that the breakthrough that makes the human brain different is that of language. It is powerful because it allows us to learn from other people’s imaginations and allows ideas to accumulate across generations. And we know that language emerges in the human brain through a hardwired curriculum to learn it that repurposes older mentalizing neocortical areas into language areas.
有了这些知识，我们现在可以转向我们祖先早期人类的真实故事。我们可以问：为什么祖先人类被赋予了这种奇怪而特殊的交流方式是什么？或者也许更重要的是：为什么许多其他聪明的动物——黑猩猩、鸟类、鲸鱼——没有这种奇怪而特殊的交流方式？大多数与语言一样强大的进化技巧都是由多个谱系独立发现的；眼睛、翅膀和多细胞性都独立进化了多次。事实上，模拟甚至心理化似乎都是沿着其他谱系独立进化的（鸟类表现出模拟的迹象，灵长类以外的其他哺乳动物表现出心理理论的迹象）。然而，至少就我们所知，语言只出现过一次。为什么？
With this knowledge, we can now turn to the actual story of our ancestral early humans. We can ask: Why were ancestral humans endowed with this odd and specific form of communication? Or perhaps more important: Why were the many other smart animals—chimps, birds, whales—not endowed with this odd and specific form of communication? Most evolutionary tricks that are as powerful as language are independently found by multiple lineages; eyes, wings, and multicellularity all independently evolved multiple times. Indeed, simulation and perhaps even mentalizing seem to have independently evolved along other lineages (birds show signs of simulation, and other mammals outside of just primates show hints of theory of mind). And yet language, at least as far as we know, has emerged only once. Why?
21
21
完美风暴
The Perfect Storm
假设你把目前发现的所有我们祖先的成年化石头骨都拿出来，用碳测定年代（这可以告诉你他们大约在多久前死去），然后测量头骨内部空间的大小（这是衡量他们大脑大小的一个很好的指标）。然后假设你绘制了这些祖先大脑随时间变化的图表。科学家已经这样做了，你得到的是图 21.1。
SUPPOSE YOU TOOK all the presently discovered adult fossilized skulls of our ancestors, carbon-dated them (which tells you approximately how long ago they died), and then measured the size of the spaces inside their skulls (a good proxy for the size of their brains). And then suppose you graphed the size of these ancestral brains over time. Scientists have done this, and what you get is figure 21.1.
大约 700 万年前，我们与黑猩猩分道扬镳，直到大约 250 万年前，我们的大脑大小基本保持不变，这时发生了一件神秘而戏剧性的事情。人类的大脑迅速增大了三倍多，成为地球上最大的大脑之一。用神经学家约翰·英格拉姆的话来说，200 多万年前，某种神秘的力量引发了“大脑失控生长。”
We diverged from chimpanzees around seven million years ago, and brains stayed largely the same size until around two and a half million years ago, at which point something mysterious and dramatic happened. The human brain rapidly became over three times larger and earned its place as one of the largest brains on Earth. In the words of the neurologist John Ingram, some mysterious force more than two million years ago triggered a “runaway growth of the brain.”
究竟为什么会发生这种情况是古人类学中一个悬而未决的问题。我们只有零星的考古线索：少量古代工具、篝火的痕迹、祖先的头骨碎片、猎杀尸体的残骸、DNA 片段、洞穴壁画和史前珠宝的碎片。我们对事件时间线的理解随着每次新的考古发现而改变。已知最早的 [X] 证据只是最早的证据，直到一位雄心勃勃的新古人类学家发现更早的样本。但尽管时间线不断变化，科学家仍然有足够的证据来重建我们一般故事的基本内容。一切始于一片垂死的森林。
Why exactly this happened is an outstanding question in paleoanthropology. We have only sparse archaeological clues: smatterings of ancient tools, hints of campfires, ancestral skull fragments, remnants of hunted carcasses, snippets of DNA, cave paintings, and broken pieces of prehistoric jewelry. Our understanding of the timeline of events changes with each new archaeological finding. The earliest known evidence of [X] is only the earliest until a new ambitious paleoanthropologist uncovers an even earlier sample. But despite this shifting timeline, there is still more than enough evidence for scientists to reconstruct the basics of our general story. It begins with a dying forest.
东区猿猴
The East Side Ape
直到一千万年前，东非还是一片树木繁茂的绿洲，一望无际的密林中，我们的祖先可以在这里采摘果实，躲避捕食者。后来，不断移动的板块开始挤压大块的土地，形成了新的地形和山脉，一直延伸到今天的埃塞俄比亚。今天，这一地区被称为东非大裂谷。
Until ten million years ago, eastern Africa was an arboreal oasis, endless acres of densely packed trees in which our ancestors could forage fruit and hide from predators. Then shifting tectonic plates began squeezing huge chunks of earth together, constructing new terrain and mountain ranges down the length of today’s Ethiopia. This region is today named the Great Rift Valley.
这些新的山脉和山谷扰乱了海洋水分的充足供应森林赖以生存的气候正是从那时开始形成的。现在东非的气候开始形成；随着森林逐渐消亡，土地变成了一片充满树木斑块和广阔草原的地形。这是最终成为今天非洲大草原的转变的开始。没有了茂密的森林，我们祖先采食热带水果和坚果的生态位开始慢慢消失。
These new mountains and valleys disrupted the bountiful supply of ocean moisture on which the forest depended. This was when the familiar climate that currently makes up eastern Africa began to take shape; as the forest slowly died, the land transformed into a terrain filled with patterned mosaics of tree patches and vast open grasslands. This was the beginning of the transformation that would eventually become today’s African savannah. Without thick forests, our ancestors’ ecological niche of foraging tropical fruit and nuts began to slowly disappear.
大约六百万年前的某个时候，这些新的山脉变成了如此庞大，以至于它们将大裂谷两侧的猿类祖先分开，将它们分成两个不同的谱系。在西侧，在森林茂密、基本没有变化的环境中，该谱系也保持了类似的变化，成为了今天的黑猩猩。然而，在山脉的东侧，在树木枯死、草原逐渐开阔的环境中，进化压力开始发生变化。正是这个谱系最终会变成人类。
At some point around six million years ago, these new mountains became so sprawling that they separated the ape ancestors on each side of the Great Rift Valley, splitting them into two separate lineages. On the western side, in an environment still rich with forests and largely unchanged, the lineage remained similarly unchanged and became today’s chimpanzees. On the eastern side of the mountains, however, in an environment of dying trees and progressively more open grasslands, evolutionary pressures began tinkering. It was this lineage that would eventually become human.
时间快进到四百万年前：这些东部猿类看起来与东非大裂谷另一边的黑猩猩表亲大体相同，只是他们现在用两条腿走路。一条腿而不是四条腿。关于为什么双足行走能帮助我们的祖先在气候变化中生存，有很多理论——也许它减少了暴露在烈日下的体表面积；也许它提高了我们的视点，让我们能够俯瞰大草原的高草；也许它帮助我们涉过浅水去获取海鲜。
Fast-forward to four million years ago: These East Side apes would have looked mostly the same as their chimpanzee cousins on the other side of the Great Rift Valley except that they were now walking on two legs instead of four. There are many theories as to why bipedalism helped our ancestors survive the changing climate—perhaps it reduced surface area exposed to the scorching sun; perhaps it elevated our eye position so we could look over the tall grass of the savannah; perhaps it helped us wade through shallow water to get seafood.
无论双足行走是为了适应什么，它都不需要额外的脑力。大约四百万年前我们直立行走的祖先的化石揭示了大脑的大小现代黑猩猩的智力水平。没有证据表明这些祖先比现代黑猩猩更聪明；没有额外的工具使用或聪明考古记录中已经发现了一些技巧。我们的祖先本质上是直立行走的黑猩猩。
Whatever bipedalism was an adaptation for, it required no extra brainpower. Fossils of our upright-walking ancestors from around four million years ago reveal a brain still the size of a modern chimpanzee’s. There is no evidence these ancestors were any smarter; no extra tool use or clever tricks have been found in the archaeological record. Our ancestors were, in essence, upright-walking chimpanzees.
图 21.2：东边的猿猴和西边的猿猴
Figure 21.2: The east side apes and the west side apes
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
直立人与人类的崛起
Homo Erectus and the Rise of Humans
250 万年前，新的非洲大草原上已经遍布大量大型食草哺乳动物；祖先大象、斑马、长颈鹿和猪在这里游荡和吃草。大草原还成为各种肉食哺乳动物的家园，包括常见的猎手如豹子、狮子和鬣狗，以及现已灭绝的动物，如剑齿虎和巨大的水獭类野兽。
By two and a half million years ago, the new African savannah had become heavily populated with massive herbivorous mammals; ancestral elephants, zebras, giraffes, and hogs wandered and grazed. The savannah also became home to diverse populations of carnivorous mammals, familiar hunters like leopards, lions, and hyenas along with a cast of now extinct animals like saber-toothed tigers and gargantuan otter-like beasts.
在这个嘈杂的大型哺乳动物动物园中，有一只不起眼的猿猴，它被迫离开舒适的森林栖息地。这只不起眼的猿猴——我们的祖先——可能正在这个充满大型食草动物和食肉动物大军的生态系统中寻找新的生存空间。
And amid this cacophonous zoo of large mammals was a humble ape who had been displaced from its comfortable forest habitat. And this humble ape—our ancestor—would have been searching for a new survival niche in this ecosystem brimming with armies of giant herbivores and carnivorous hunters.
我们的祖先似乎落入了最初的生态位吃尸体。我们的祖先开始转向吃肉。黑猩猩的饮食中只有大约 10% 来自肉类，而有证据表明，这些早期人类的饮食中多达 30% 来自肉类来自肉。
The initial niche our ancestors seemed to fall into was scavenging carcasses. Our ancestors began shifting toward eating meat. Only about 10 percent of the diet of a chimpanzee comes from meat, while evidence suggests that as much as 30 percent of the diet of these early humans came from meat.
我们从他们留下的工具和骨头痕迹推断出这种拾荒生活方式。这些祖先发明了石器，似乎专门用于处理尸体的肉和骨头。这些工具被称为“奥尔杜威工具”，以它们的发现地点（坦桑尼亚的奥杜威峡谷）命名。
We infer this scavenging lifestyle from the tools and bone markings they left behind. These ancestors invented stone tools that seemed to be used specifically for processing the meat and bones of carcasses. These tools are referred to as “Oldowan tools” after the location where they were discovered (Olduvai Gorge in Tanzania).
我们的祖先分三个步骤制造这些工具：（1）他们找到由坚硬的岩石制成的锤石；（2）他们找到由更易碎的石英、黑曜石或玄武岩制成的岩芯；（3）他们用锤石敲击岩芯，产生多个锋利的薄片和一把尖头砍刀。
Our ancestors constructed these tools in three steps: (1) They found a hammerstone made of hard rock; (2) they found a core made of more fragile quartz, obsidian, or basalt; (3) they smashed the hammerstone against the core to produce multiple sharp flakes and a pointed chopper.
猿类的身体不适合食用大量肉类；虽然狮子可以用巨大的牙齿切开厚厚的兽皮，把肉从骨头上撕下来，但我们的祖先没有这样的天然工具。所以我们的祖先发明了人造工具。石片可以切开兽皮，切下肉，石斧可以砸开骨头，获取营养丰富的骨髓。
Ape bodies aren’t adapted for consuming large quantities of meat; while lions can use their massive teeth to slice through thick hides and rip meat off bones, our ancestors had no such natural tools. So our ancestors invented artificial tools. Stone flakes could slice through hides and cut away meat, and stone choppers could smash open bones to access nutritious marrow.
图 21.3：奥杜威工具的制造
Figure 21.3: Manufacture of Oldowan tools
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
快进五十万年，我们东非的祖先进化成了一种名为直立人的物种，意思是“直立的人”（这个名字很傻，因为我们的祖先在直立人出现之前就已经直立行走了）。直立人表示人类的属，而直立人表示人类的特定物种。直立人的出现标志着人类进化的一个转折点。早期的人类是胆小的秃鹫，而直立人 是一种顶级捕食者。
Fast-forward five hundred thousand years, and our ancestors in eastern Africa had evolved into a species called Homo erectus, meaning “upright man” (which is a silly name, since our ancestors were walking upright well before Homo erectus). Homo denotes the genus of humans, and erectus denotes a specific species of human. The emergence of Homo erectus marked a turning point in human evolution. While earlier humans were timid vultures, Homo erectus was an apex predator.
直立人变成了超级肉食动物，他们的饮食几乎荒谬的 85% 的肉。直立人可能非常成功，以至于他们取代了当地的竞争对手；在直立人出现的时候，非洲大草原上的许多其他食肉动物开始灭绝。
Homo erectus became a hypercarnivore, consuming a diet that was an almost absurd 85 percent meat. Homo erectus may have been so successful that they displaced their local competitors; around the time Homo erectus appeared, many of the other carnivores in the African savannah began to go extinct.
直立人有许多身体上的适应性变化，揭示了他们捕食的生活方式，而我们现代人都继承了这些变化。最值得注意的是，直立人的大脑比一百万年前我们直立行走的黑猩猩祖先的大脑大两倍。更大的大脑至少有一个好处，那就是可以制造更好的工具：直立人发明了一种新型锋利的石手斧。他们的肩膀和躯干变得特别适合投掷。成年黑猩猩比人类强壮得多，但由于肩膀和躯干僵硬，它只能以每小时 20 英里的速度投掷抛射物。一个相对弱小的青春期前人类几乎可以投掷一个球速度快三倍。我们通过一套独特的调整来做到这一点，这让我们能够增加肩部的张力然后猛地挥动手臂。投掷石块或长矛可能是抵御掠食者、从其他食肉动物那里偷肉，甚至主动猎杀羚羊和猪的一种手段。
Homo erectus had numerous physical adaptations that reveal their predator lifestyle, all of which we modern humans have inherited. Most notable, H. erectus had a brain that was twice the size of our ancestral upright-walking-chimpanzee-like ancestor’s from a million years prior. At least one benefit of this bigger brain was better tools: H. erectus invented a new class of sharp stone hand axes. Their shoulders and torsos became uniquely adapted for throwing. An adult chimp is far stronger than a human, and yet with its stiff shoulders and torso, it can throw a projectile at a speed of only about twenty miles per hour. A relatively puny preadolescent human can throw a ball almost three times as fast. We do this with a unique set of adjustments that allow us to build up tension in our shoulders and then snap and fling our arms. Throwing rocks or spears may have been a trick for defending against predators, stealing meat from other carnivores, or even actively hunting antelopes and hogs.
直立人也进化出了耐力奔跑的适应能力。腿变长，脚变得更拱，皮肤变得无毛，汗腺增生。直立人和现代人都有一种独特的降温方法——其他哺乳动物通过喘气来降低体温，而现代人则通过出汗。这些特征使我们的祖先在炎热的大草原上长途跋涉时身体保持凉爽。虽然现代人并不是跑得最快的生物，但我们实际上是动物界中耐力最好的跑步者之一；即使是猎豹也无法一次跑完 26 英里的马拉松。一些人认为直立人使用了一种称为持久狩猎的技巧——追逐猎物，直到猎物累得无法再走下去。这正是南部非洲喀拉哈里沙漠的现代狩猎采集者所使用的技巧。
Homo erectus also evolved adaptations for endurance running. Legs elongated, feet became more arched, skin became hairless, and sweat glands proliferated. Both Homo erectus and modern humans have a peculiar method of cooling down—while other mammals pant to lower their body temperature, modern humans sweat. These traits would have kept our ancestors’ bodies cool while they were trekking long distances in the hot savannah. While modern humans are hardly the fastest creatures, we are actually some of the best endurance runners in the animal kingdom; even a cheetah couldn’t run a twenty-six-mile marathon in one go. Some believe H. erectus used a technique called persistence hunting—chasing prey until it was simply too tired to go any farther. This is exactly the technique used by modern hunter-gatherers in the Kalahari Desert of southern Africa.
直立人的嘴巴和内脏都缩小了。人类相对于猿类的熟悉面孔主要是由于下颌缩小，鼻子更加突出。这些变化令人困惑；由于身体和大脑更大，直立人需要更多的能量，因此需要更强的下颌和更长的消化道来消耗更多的食物。20 世纪 90 年代，灵长类动物学家理查德·兰厄姆提出了一个理论来解释这一现象：直立人 一定发明了烹饪。
The mouths and guts of Homo erectus shrank. The familiar face of a human relative to an ape is mostly a consequence of a shrunken jaw, which makes the nose more prominent. These changes are perplexing; with a bigger body and brain, Homo erectus would have needed more energy and thus stronger jaws and longer digestive tracts for consuming more food. In the 1990s, the primatologist Richard Wrangham proposed a theory to explain this: H. erectus must have invented cooking.
当肉或蔬菜被烹饪时，较难消化的细胞结构被分解成能量更丰富的化学物质。烹饪使动物能够多吸收 30% 的营养，并减少消耗消化所需的时间和能量。事实上，现代人类特别依赖烹饪来消化。每种人类文化都使用烹饪，而那些试图吃完全生食（无论是生肉还是生蔬菜）的人，会长期缺乏能量，而且50%的人会暂时不孕。
When meat or vegetables are cooked, harder-to-digest cellular structures are broken down into more energy-rich chemicals. Cooking enables animals to absorb 30 percent more nutrients and spend less time and energy digesting. In fact, modern humans are uniquely reliant on cooking for digestion. Every human culture uses cooking, and humans who attempt to eat fully raw diets, whether raw meat or raw vegetables, have chronic energy shortages, and over 50 percent become temporarily infertile.
人类控制使用火的第一个证据可以追溯到直立人出现的时候，当时我们发现了烧焦的骨头和古代洞穴中的灰烬。直立人可能故意用石头砸来生火，或者他们可能使用天然的森林火灾，捡起燃烧的木棍。无论哪种方式，食用熟肉都会火提供了独特的热量盈余，可以无差别地用于更大的大脑。正如许多宗教和文化所神话的那样，火的馈赠可能让我们的祖先走上了不同的轨迹。
The first evidence of controlled use of fire by humans dates to around the time Homo erectus came on the scene, where we find hints of charred bones and ash in ancient caves. Homo erectus may have deliberately created fire by smashing stones together, or they may have used natural forest fires, picking up flaming sticks. Either way, consuming cooked meat would have offered a unique caloric surplus that could be indiscriminately spent on larger brains. As many religions and cultures have mythologized, it may have been the gift of fire that put our ancestors on a different trajectory.
随着直立人大脑的扩大，一个新的问题出现了：大脑体积大，难以通过产道。人类的双足行走会进一步加剧这一问题，因为直立需要更窄的臀部。这就是人类学家谢伍德·沃什伯恩所说的“产科困境”。人类解决这个问题的办法是早产。新生母牛出生后数小时内就能走路，新生猕猴出生后两个月内就能走路，但人类新生儿往往一年内都无法独立行走 出生后。人类不是在准备好出生时出生，而是在大脑达到能够通过产道的最大尺寸时出生。
As the brain of Homo erectus expanded, a new problem would have emerged: Big brains are hard to fit through birth canals. Human bipedalism would have further exacerbated this problem, as standing upright requires narrower hips. This is what the anthropologist Sherwood Washburn calls the “obstetric dilemma.” The human solution to this is premature birthing. A newborn cow can walk within hours of being born, and a newborn macaque monkey can walk within two months, but newborn humans often can’t walk independently for up to a year after they are born. Humans are born not when they are ready to be born, but when their brains hit the maximum size that can fit through the birth canal.
除了出生时大脑发育早之外，人类大脑发育的另一个独特之处是人类大脑需要很长时间才能达到成年时的大小。人类大脑需要十二年才能达到成年时的大小，这在动物界最聪明、脑容量最大的动物中创下了纪录。
Another unique feature of human brain development, in addition to how premature brains are at birth, is how long it takes for human brains to reach their full adult size. Setting a record among even the smartest and biggest-brained animals in the animal kingdom, it takes a human brain twelve years before it has reached its full adult size.
早产和儿童大脑发育期延长给直立人带来了改变养育方式的压力。黑猩猩新生儿大部分完全由母亲抚养。但考虑到人类婴儿的早产率以及他们需要多长时间的支持，这对直立人母亲来说要困难得多。许多古人类学家认为，这使得直立人的群体动态从黑猩猩的滥交转变为我们所见的（大多数）一夫一妻制的配偶关系在当今人类社会中。有证据表明，直立人父亲在照顾孩子方面发挥了积极作用，并且这种配对关系持续了很长一段时间。
Premature birthing and an extended period of childhood brain growth put pressure on H. erectus to change its parenting style. Chimpanzee newborns are, for the most part, entirely raised by their mothers. But this would have been much more difficult for a Homo erectus mother given how premature human infants are born, and how long they need support. Many paleoanthropologists believed this shifted Homo erectus group dynamics away from the promiscuous mating of chimpanzees to the (mostly) monogamous pair-bonding we see in today’s human societies. Evidence suggests that Homo erectus fathers took an active role in caring for their children and that these pairings persisted for long periods.
“祖母式” 可能也出现在直立人身上。地球上只有两种哺乳动物生下的雌性直到死亡才具有生殖能力：虎鲸和人类。人类女性会经历更年期，并在之后存活很多年。一种理论认为，更年期的进化是为了促使祖母将注意力从抚养自己的孩子转移到抚养孩子的孩子上。祖母式在各个文化中都有所体现，甚至在当今狩猎采集社会。
“Grandmothering” may also have emerged in Homo erectus. Only two mammals on Earth produce females that are not reproductively capable until death: orcas and humans. Human females go through menopause and live for many years afterward. One theory is that menopause evolved to push grandmothers to shift their focus from rearing their own children to supporting their children’s children. Grandmothering is seen across cultures, even in present-day hunter-gatherer societies.
物种 SPECIES |
出生时成人大脑大小的百分比 PERCENT OF ADULT BRAIN SIZE AT BIRTH |
|
人类 Human |
28% 28 percent |
12年 12 years |
黑猩猩 Chimpanzee |
36% 36 percent |
6 年 6 years |
猕猴 Macaque |
70% 70 percent |
3 年 3 years |
直立人是我们吃肉、使用石器、（可能）使用火、早产、（大部分）一夫一妻制、祖母式生活、无毛、流汗、大脑发达的祖先。当然，价值百万美元的问题是，直立人会说话吗？
Homo erectus was our meat-eating, stone-tool-using, (possibly) fire-wielding, premature-birthing, (mostly) monogamous, grandmothering, hairless, sweating, big-brained ancestor. The million-dollar question is, of course, did Homo erectus speak?
华莱士的问题
Wallace’s Problem
早在达尔文发现进化论之前，人们就在思考语言的起源。柏拉图考虑过这个问题。《圣经》描述了这个问题。从让-雅克·卢梭到托马斯·霍布斯，许多启蒙运动的知识分子都曾思考过人类的自然状态，并对此进行推测。
Long before Darwin discovered evolution, people were pondering the origins of language. Plato considered it. The Bible describes it. Many of the Enlightenment intellectuals who contemplated humankind’s state of nature, from Jean-Jacques Rousseau to Thomas Hobbes, speculated about it.
因此，达尔文出版《物种起源》后，人们对语言的进化起源的猜测如潮水般涌来，这并不令人意外，这一次是在达尔文自然选择理论的背景下。1866 年，也就是达尔文出版这本书的七年后，法国科学院对这些未经证实的猜测数量感到厌烦，禁止出版关于人类语言的起源。
And so it is unsurprising that immediately after Darwin published his Origin of Species, there was a tidal wave of speculations on the evolutionary origins of language, this time within the context of Darwin’s theory of natural selection. In 1866, just seven years after Darwin’s book, the French Academy of Sciences was so fed up with the quantity of these unsubstantiated speculations that they banned publications about the origin of human languages.
阿尔弗雷德·华莱士被许多人视为进化论的共同创始人之一，他曾承认进化论可能永远无法解释语言，甚至援引上帝的概念来解释语言。达尔文对华莱士的退缩感到非常懊恼，于是给他写了一封信，怒气冲冲地说：“我希望你没有把你的进化论扼杀得太彻底我自己和我的孩子。”进化论的共同创始人之一拒绝了进化论的解释，这一让步成为了如此臭名昭著的让步，以至于寻找语言进化解释的问题被俗称为“华莱士问题”。
Alfred Wallace, whom many consider one of the cofounders of the theory of evolution, famously conceded that evolution might never be able to explain language and even invoked the notion of God to explain it. So chagrined by Wallace’s retreat, Darwin wrote him a letter, fuming: “I hope that you have not murdered too completely your own and my child.” This rejection of an evolutionary explanation by one of the cofounders of the theory of evolution became such an infamous concession that the problem of finding an evolutionary explanation for language has been colloquially dubbed “Wallace’s problem.”
在过去的 150 年里，新的推测不断涌现，以符合新的证据，但并没有太大的变化——人类何时首次使用语言以及语言进化过程中经历了哪些渐进阶段仍然是人类学、语言学和进化心理学中最具争议的两个问题。有些人甚至认为语言的起源是“最难的所有科学领域都存在的问题。”
In the past one hundred fifty years, new speculations have emerged to align with new evidence, but not much has changed—when humans first used language and what incremental stages occurred in the evolution of language are still two of the most controversial questions across anthropology, linguistics, and evolutionary psychology. Some have even gone so far as to suggest that the origin of language is “the hardest problem in all of science.”
回答这些问题之所以如此困难，部分原因在于，目前还没有现存物种只具有少量语言的例子。相反，有些非人类灵长类动物没有自然产生的语言，而智人却有语言。如果尼安德特人或直立人成员活到了今天，我们可能会对语言的出现过程有更多的线索。但今天活着的所有人类都来自大约十万年前的共同祖先。我们现存的最近亲属是黑猩猩，我们与黑猩猩有共同的祖先，生活在七百万年前。这两个时期之间的进化空洞使我们没有任何现存物种可以解读语言进化的中间阶段。
Part of what makes answering these questions so difficult is that there are no examples of living species with only a little bit of language. Instead, there are nonhuman primates with no naturally occurring language and Homo sapiens with language. If any Neanderthals or members of Homo erectus had survived to this day, we might have far more clues as to the process by which language emerged. But all humans alive today descended from a common ancestor around one hundred thousand years ago. Our nearest living cousin is the chimp, with whom we share a common ancestor who lived over seven million years ago. The evolutionary cavern between these periods leaves us without any living species from which to decipher the intermediary stages of language evolution.
考古记录只给我们提供了两个不容置疑的里程碑，所有语言进化理论都必须面对这两个里程碑。首先，化石告诉我们，我们祖先的喉部和声带直到大约五十万年前才适应有声语言。这一特征并非智人独有——尼安德特人也有准备好语言的声带。这意味着，如果语言在此之前就存在，那么它应该主要是手势语言，或者是一种不太复杂的口头语言。其次，大量证据表明语言至少在十万年前就存在了。符号学的一致证据——以虚构的雕塑、抽象的洞穴艺术和非功能性珠宝为衡量标准——出现在大约十万年前；许多人认为，这种符号学只有在语言中才有可能。此外，所有现代人类都表现出同等的语言能力，这表明我们十万年前的共同祖先几乎肯定说着同样复杂的语言。
The archaeological record gives us only two indisputable milestones that all theories of language evolution must contend with. First, fossils tell us that the larynx and vocal cords of our ancestors were not adapted to vocal language until about five hundred thousand years ago. This trait was not unique to Homo sapiens—Homo neanderthalensis too had language-ready vocal cords. This means that if language existed before this time, it would have been primarily gestural or would have been a less complex verbal language. Second, substantial evidence suggests that language existed by at least one hundred thousand years ago. Consistent evidence of symbology—as measured by fictional sculptures, abstract cave art, and nonfunctional jewelry—shows up at around one hundred thousand years ago; many argue that such symbology would only have been possible with language. Further, all modern humans exhibit equal language proficiencies, suggesting that our common ancestor from one hundred thousand years ago almost definitely spoke an equally complex language.
秉承这些里程碑，现代语言进化的故事涵盖了所有可能性。有些人认为基本的原始语言语言出现于 250 万年前，直立人出现之前最早的人类身上；另一些人认为语言出现于 10 万年前，只有智人才有。一些人认为语言的进化是渐进的；另一些人认为语言的进化是迅速且突然发生的。一些人认为语言是从手势开始的；另一些人认为语言是从口头开始的。
Adhering to these milestones, the modern stories of language evolution run the entire gamut of possibilities. Some argue basic protolanguages emerged two and a half million years ago with the very first humans before Homo erectus; others argue that it emerged as late as one hundred thousand years ago uniquely in Homo sapiens. Some argue that language evolution was gradual; others say it occurred rapidly and all at once. Some argue that language began gesturally; others say that it began verbally.
图 21.4：重建语言进化时间线的线索
Figure 21.4: Clues for reconstructing the timeline of language evolution
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
这些争论常常以新的形式重述旧观点；从很多方面来看，今天关于语言进化的故事与 150 多年前法国禁止讨论语言进化的故事一样，都是推测性的。但从其他方面来看，情况已经不同了。我们对行为、大脑和考古记录有了更深入的了解。也许最重要的是，我们对进化的机制有了更深入的了解，而正是在这里，我们找到了语言起源的最大线索。
These debates often restate old ideas in new forms; in many ways, today’s stories of language evolution are just as speculative as they were when the French banned discussions of it over one hundred fifty years ago. But in other ways, things are different. We have a far greater understanding of behavior, brains, and the archaeological record. And perhaps most important, we have a far greater understanding of the machinations of evolution, and it is here where we find our greatest clue to the origin of language.
利他主义者
The Altruists
直觉上，语言的进化应该出于与其他有用的进化适应相同的原因。以眼睛为例。如果人类如果人类 B 的眼睛比人类 A 的眼睛稍微好一些，那么人类 A 成功狩猎和交配的概率就会更高。因此，随着时间的推移，眼睛更好的基因应该会在人群中传播。
It is intuitive to argue that language should have evolved for the same reason as any other useful evolutionary adaptation. Take the eye. If human A had slightly better eyes than human B, then human A had a higher probability of successfully hunting and mating. Hence, over time, the better-eyes gene should propagate through the population.
然而，语言与眼睛却存在着重大区别。语言并不像眼睛那样直接使人受益；只有当其他人以有用的方式使用语言时，语言才能使人受益。
There is a crucial difference, however, with language. Language doesn’t directly benefit an individual the way eyes do; it benefits individuals only if others are using language with them in a useful way.
好吧，也许适用于个人的进化逻辑也适用于群体：如果人类 A 组进化出了一点语言，而人类 B 组没有语言，那么人类 A 组会生存得更好，因此语言的任何进步改进都会被选择。
Well, perhaps the same evolutionary logic that applies to individuals might apply to groups: If group A of humans evolved a little bit of language, and group B of humans had no language, then group A would survive better, hence any progressive improvements to language would be selected for.
这种推理引发了进化生物学家所谓的“群体选择”。群体选择是利他行为的直观解释。如果一种行为降低了个体的生殖适应性，但提高了另一个人的生殖适应性，那么这种行为就是利他行为。语言的许多好处，从定义上讲，都是利他行为——它是分享食物位置、警告危险区域、明确教授使用工具。简单的群体选择论证表明，语言等利他行为之所以进化，是因为进化有利于物种的生存，因此个体愿意为了更大的利益做出牺牲。
This type of reasoning invokes what evolutionary biologists call “group selection.” Group selection is an intuitive explanation for altruistic behaviors. A behavior is altruistic if it decreases an individual’s reproductive fitness but increases another’s reproductive fitness. Many of the benefits of language are, by definition, altruistic—it is the sharing of food locations, warning of dangerous areas, the explicit teaching of tool use. The simple group selection argument suggests that altruistic behaviors such as language evolve because evolution favors the survival of the species, and individuals are therefore willing to make sacrifices for the greater good.
虽然许多现代生物学家都同意这种群体层面的效应在进化过程中确实存在，但这些群体层面的效应远比支持物种生存的简单特征选择更加微妙和复杂。进化不是这样进行的。问题在于基因不会自发出现在群体中，而是出现在个体中。
While many modern biologists agree that such group-level effects do occur in evolution, these group-level effects are far more nuanced and complex than the simple selection of traits that support the survival of the species. Evolution does not work this way. The problem is that genes do not spontaneously appear in groups, they appear in individuals.
假设 A 组中 10% 的人是利他主义者——他们自由地分享信息，教别人如何使用工具，并透露食物的位置。假设另外 90% 的人不是利他主义者——他们不分享食物的位置，也不花时间教别人如何使用工具。为什么这群利他主义者会过得更好？一个乐于接受这些学习但不给予任何回报的吃白食者难道不会比利他主义者活得更好吗？
Suppose 10 percent of group A is altruistic—they freely share information, teach others how to use tools, and reveal the locations of food. And assume the other 90 percent is not altruistic—they don’t share locations of food or spend time teaching tool use. Why would this subgroup of altruists fare any better? Wouldn’t a freeloader who was happy to accept these learnings but gave nothing in return survive better than the altruists?
利他主义并不是生物学家所说的进化稳定策略。侵犯、欺骗和搭便车的策略似乎更有利于个体基因的生存。
Altruism is not what biologists call an evolutionarily stable strategy. The strategy of violating, cheating, and freeloading seems to better serve the survival of one’s individual genes.
那么，按照这个论点，动物界是如何出现合作行为的呢？事实证明，动物的大多数群体行为都不是利他主义；它们是互利的安排，对所有参与者来说都是净收益。鱼群游动是因为它对它们所有人都有好处，而这种运动实际上最好的解释是边缘的鱼都在争先恐后地进入中心那里最安全。角马会聚在一起，因为它们聚在一起时更安全。
But then, by this argument, how do any cooperative behaviors emerge in the animal kingdom? It turns out most group behaviors in animals aren’t altruistic; they are mutually beneficial arrangements that are net-positive for all participants. Fish swim in shoals because it benefits all of them, and the movements are actually best explained by the fish on the edges all fighting to get into the center where it is safest. Wildebeests band together because they are all safer when they are in a group.
在所有这些情况下，背叛只会伤害到你自己。决定离开鱼群独自游动的鱼会首先被吃掉。角马也是一样。但语言不是这样；语言上的背叛——直接撒谎或隐瞒信息——对个人有很多好处。而撒谎者和骗子的存在会破坏语言的价值。在一个每个人都用语言互相撒谎的群体中，那些不会说语言、不受谎言影响的人实际上可能比那些会说语言的人生存得更好。因此，语言的存在为背叛者创造了一个利基市场，从而消除了语言的原始价值。那么，语言如何在一个群体中传播和延续呢？
In all these situations, defecting hurts only yourself. A fish that decides to leave the shoal and swim on its own will be the first to be eaten. Same for a wildebeest. But language is not like this; defecting in language—directly lying or withholding information—has many benefits to an individual. And the presence of liars and cheaters defeats the value of language. In a group where everyone is lying to each other with words, those who spoke no language and were immune to the lies might in fact survive better than those with language. So the presence of language creates a niche for defectors, which eliminates the original value of language. How, then, could language ever propagate and persist within a group?
这样一来，人类大脑进化的第五大突破——语言——与本书记载的其他突破都不同。引导、强化、模拟和心智化都是适应性行为，这些适应性行为显然对它们开始出现的任何个体生物都有好处，因此它们繁衍的进化机制是直截了当的。然而，语言只有在一群人使用时才有价值。因此，一定有更微妙的进化机制在起作用。
In this way, the fifth breakthrough in the evolution of the human brain—language—is unlike any other breakthrough chronicled in this book. Steering, reinforcing, simulating, and mentalizing were adaptations that clearly benefited any individual organisms in which they began to emerge, and thus the evolutionary machinations by which they propagated are straightforward. Language, however, is only valuable if a group of individuals are using it. And so more nuanced evolutionary machinations must have been at work.
动物界有两种类型的利他主义。第一种称为亲属选择。亲属选择是指个体为了直接亲属的利益而做出个人牺牲。基因有两种方式可以持续存在：提高宿主的生存机会或帮助宿主的兄弟姐妹和孩子生存。孩子和兄弟姐妹都有 50% 的机会分享你的一个基因。孙子有 25% 的机会。堂兄弟有 12.5% 的机会。在进化压力的背景下，实际上有一个数学表达式来比较生物体对自己生命的价值与其亲属生命的价值。正如进化生物学家 JBS Haldane 的名言：“我很乐意为两个兄弟或八个堂兄弟献出生命。”这就是为什么许多鸟类、哺乳动物、鱼类和昆虫为它们的后代做出个人牺牲，而为堂兄弟和陌生人做出个人牺牲的机会却少得多。
There are two types of altruism found in the animal kingdom. The first is called kin selection. Kin selection is when individuals make personal sacrifices for the betterment of their directly related kin. A gene has two ways to persist: improve its host’s chance of survival or help the host’s siblings and children survive. A child and sibling both have a 50 percent chance of sharing one of your individual genes. A grandchild has a 25 percent chance. A cousin has a 12.5 percent chance. In the context of evolutionary pressures, there is literally a mathematical expression comparing the value an organism places on its own life relative to that of its relatives. As the evolutionary biologist J. B. S. Haldane famously quipped: “I would happily lay down my life for two brothers or eight cousins.” This is why many birds, mammals, fish, and insects make personal sacrifices for their offspring but much less so for cousins and strangers.
当我们通过这个视角重新审视其他社会生物的行为时，就会发现大多数利他行为都是亲属选择的结果。长尾黑尾猴主要在以下情况下发出警报：围绕着家庭成员。细菌彼此共享基因，因为它们是克隆体。蚁群和蜂巢展示了成千上万个体之间令人难以置信的合作和牺牲。群体选择？不，这都是亲属选择，这是因为它们独特的社会结构。一个蜂巢里只有一只蜂王，她负责整个蜂巢的繁殖。这确保了蜂巢是由姐妹和兄弟组成的。一只工蜂传播基因的最佳方式是照顾整个蜂巢和蜂王，因为根据定义，蜂王与工蜂共享大部分基因。
When we reexamine the behavior of other social creatures through this lens, it becomes clear that most altruistic behaviors are the result of kin selection. Vervet monkeys primarily make their alarm calls when they are around family members. Bacteria share genes with each other because they are clones. Ant colonies and beehives show incredible cooperation and sacrifice among tens of thousands of individuals. Group selection? No, it is all kin selection, and this works because of their unique social structure. A beehive has a single queen bee who does all the reproduction for the entire beehive. This ensures that the beehive is made up of sisters and brothers. The best way for an individual worker bee to propagate its genes is to care for the entire beehive and the queen, who, by definition, share most of its genes.
除了亲属选择之外，动物界中发现的另一种利他主义是互惠利他主义是一种行为，称为互惠利他主义。互惠利他主义相当于“你帮我，我帮你”。个体会做出今天的牺牲，以换取未来的互惠利益。我们在灵长类动物梳理毛发时已经看到了这一点——许多灵长类动物会为没有血缘关系的个体梳理毛发，而那些被梳理毛发的个体在受到攻击时更有可能跑去帮助梳理毛发的人。黑猩猩会选择性地与支持过无血缘关系的个体分享食物过去，我们曾看到，这些联盟并非无私的，而是互惠互利的：“我现在会帮助你，但下次我受到攻击时，请你保护我。”
In addition to kin selection, the other type of altruism found in the animal kingdom is called reciprocal altruism. Reciprocal altruism is the equivalent of “I’ll scratch your back if you scratch mine.” An individual will make a sacrifice today in exchange for a reciprocal benefit in the future. We saw this already in primate grooming—many primates groom unrelated individuals, and those that get groomed are more likely to run to the aid of the groomer when attacked. Chimpanzees selectively share food with unrelated individuals who have supported them in the past. These alliances are, as we saw, not selfless; they are reciprocally altruistic: “I will help you now, just please protect me next time I get attacked.”
互惠利他主义要在整个群体中成功传播，其基本特征是发现和惩罚叛逃者。没有这一点，利他行为最终会创造出搭便车的人。最常见的说法是“骗我一次，你可耻；骗我两次，我可耻。”这类动物似乎默认帮助他人，但当其他人不回报时，它们就停止利他行为。红翅黑鸟会保卫附近无亲属的邻居的巢穴，这是高度利他的行为，因为保卫巢穴是有风险的，但它们似乎期望得到回报。事实上，当这种帮助得不到回报时，黑鸟会选择性地停止帮助这些个体这对他们没有帮助。
The essential feature for reciprocal altruism to successfully propagate throughout a group is the detection and punishment of defectors. Without that, altruistic behaviors end up creating freeloaders. The most common version of this is the saying “Fool me once, shame on you; fool me twice, shame on me.” Such animals seem to default to helping others, but when others fail to reciprocate, they stop behaving altruistically. Red-winged blackbirds defend the nests of unrelated nearby neighbors, which is highly altruistic, since it is risky to defend nests, but they seem to do it with expectation of reciprocity. Indeed, when such help is not reciprocated, blackbirds selectively stop helping the individuals that did not help them.
然而，现代人类的许多行为并不完全符合亲属选择或互惠利他主义。当然，人类明显偏向自己的亲属。但人们仍然经常帮助陌生人，而不期望我们不求回报，不求回报。我们向慈善机构捐款；我们愿意为我们的同胞参战，冒着生命危险，尽管我们大多数人从未见过他们；我们参加的社会运动并不直接使我们受益，而是帮助那些我们认为处于不利地位的陌生人。想想看，如果一个人在街上看到一个迷路的、受惊的孩子，却什么也不做，那是多么奇怪。大多数人会停下来帮助一个孩子，而且不求回报。与其他动物相比，人类对不相干的陌生人是最无私的。
Much behavior of modern humans, however, doesn’t fit cleanly into kin selection or reciprocal altruism. Sure, humans are clearly biased toward their own kin. But people still regularly help strangers without expecting anything in return. We donate to charity; we are willing to go to war and risk our lives for our fellow citizens, most of whom we’ve never met; and we take part in social movements that don’t directly benefit us but help strangers we feel have been disadvantaged. Think about how weird it would be for a human to see a lost and scared child on the street and just do nothing. Most humans would stop to help a child and do so without expecting any reciprocity in return. Humans are, relative to other animals, by far the most altruistic to unrelated strangers.
当然，人类也是最残忍的物种之一。只有人类会做出令人难以置信的个人牺牲来将痛苦和苦难强加给他人。只有人类会进行种族灭绝。只有人类会憎恨整个群体。
Of course, humans are also one of the cruelest species. Only humans will make incredible personal sacrifices to impose pain and suffering on others. Only humans commit genocide. Only humans hate entire groups of people.
这一悖论并非偶然发生；我们的语言、无与伦比的利他主义和无与伦比的残忍在进化过程中同时出现并非巧合；事实上，这三者只是同一进化反馈回路的不同特征，而进化正是在人类大脑进化的漫长旅程中完成了这一反馈回路的画龙点睛之笔。
This paradox is not a random happenstance; it is not a coincidence that our language, our unparalleled altruism, and our unmatched cruelty all emerged together in evolution; all three were, in fact, merely different features of the same evolutionary feedback loop, one from which evolution made its finishing touches in the long journey of human brain evolution.
让我们回到直立人，看看这一切是如何结合在一起的。
Let’s return to Homo erectus and see how all this comes together.
人类蜂巢思维的出现
The Emergence of the Human Hive Mind
虽然我们永远无法确定，但证据支持直立人说的是原始语言的观点。他们可能没有说出语法丰富的短语——他们的声带只能发出很窄范围的辅音和元音（因此原始语言中有proto）。但直立人可能有能力分配陈述性标签，甚至可能使用一些简化的语法。他们的斧头状工具制造起来很复杂，但却流传了数千代；如果没有至少一些共同注意的机制，这种类型的复制是很难想象的以及语言教学。尽管它们体质虚弱、没有爪子、行动相对缓慢，但它们作为食肉动物取得了令人难以置信的成功，这表明它们具备一定程度的合作和协调能力，而如果没有语言，这种能力也不太可能实现。
While we will never know for sure, the evidence tips in favor of the idea that Homo erectus spoke a protolanguage. They may not have been uttering rich grammatical phrases—their vocal cords could make only a narrow range of consonant and vowel sounds (hence the proto in protolanguage). But Homo erectus probably had the ability to assign declarative labels and perhaps even use some simplified grammars. Their ax-like tools were complex to manufacture and yet were passed down across thousands of generations; this type of copying is hard to imagine without at least some mechanisms of joint attention and language-enabled teaching. Their incredible success as carnivores, despite being feeble, clawless, and relatively slow, suggests a degree of cooperation and coordination that is also unlikely without language.
第一句话可能来自父母与子女之间的原始对话，目的可能很简单，就是确保成功传承了先进的工具制造技术。对于其他猿类来说，工具是它们生存环境的一个有用但并非必不可少的特征。然而，对于直立人而言，制造复杂的工具是生存的必要条件。没有石手斧的直立人就像天生没有牙齿的狮子一样注定要灭亡。
The first words may have emerged from proto-conversations between parents and their children, perhaps for the simple purpose of ensuring the successful transmission of advanced tool manufacture. In other apes, tools are a useful but not essential feature of their survival niche. In H. erectus, however, the manufacture of complex tools was a requirement to survive. A Homo erectus without a stone hand ax was as doomed as a lion born without teeth.
这些原始对话可能还有其他好处，而且都不需要复杂的语法：表示在哪里可以找到食物（“浆果。家树”）、警告（“安静。危险”）和联系电话（“妈妈。在这里”）。
These proto-conversations could have had other benefits as well, none of them requiring sophisticated grammar: signaling where to find food (“Berries. Home tree”), warnings (“Quiet. Danger”), and contact calls (“Mom. Here”).
语言最初是父母与子女之间的一种把戏，这一论点有助于解释两件事。首先，语言不需要任何有争议的群体选择，只需通过常见的亲属选择即可发挥作用。选择性地使用语言帮助孩子成长为独立、成功、会使用工具的成年人，这并不比父母的其他任何形式的投资更神秘。其次，语言的学习计划在父母与子女之间共同关注和原始对话的硬性相互作用中最为突出，这暗示了语言的起源于这类关系。
The argument that language first emerged as a trick between parents and children helps explain two things. First, it requires none of the controversial group selection and can work simply through the common kin selection. Selective use of language to help rear children into independent successful tool-using adults is no more mysterious than any other form of parental investment. Second, the learning program for language is most prominent in the hardwired interplay of joint attention and proto-conversations between parents and children, suggestive of its origin in these types of relationships.
有了亲属之间的语言基础，使用语言的机会非亲属之间的交流成为可能。母亲和后代之间不再需要临时语言，而是整个群体可以共享标签。但正如我们所见，群体中不相关的个体之间共享的信息将是脆弱和不稳定的，成熟针对叛逃者和撒谎者。
With the basics of language among kin in place, the opportunity for using language among non-kin became possible. Instead of makeshift languages constructed between mothers and offspring, it would have been possible for an entire group to share in labels. But as we have seen, information shared with unrelated individuals in a group would have been tenuous and unstable, ripe for defectors and liars.
提出社会大脑假说的著名人类学家罗宾·邓巴（Robin Dunbar）在这里提出了一个巧妙的观点。我们人类天生就有一种谈论的本能？我们用语言最自然的活动是什么？嗯，我们八卦。我们常常控制不住自己；我们必须分享他人的道德违规行为，讨论关系变化，关注戏剧性事件。邓巴对此进行了测量——他窃听了公共谈话，发现多达 70% 的人类的谈话就是闲聊。对邓巴来说，这是了解语言的起源。
Here is where Robin Dunbar—the famous anthropologist who came up with the social-brain hypothesis—proposes something clever. What do we humans naturally have an instinct to talk about? What is the most natural activity we use language for? Well, we gossip. We often can’t help ourselves; we have to share moral violations of others, discuss relationship changes, keep track of dramas. Dunbar measured this—he eavesdropped on public conversations and found that as much as 70 percent of human conversation is gossip. This, to Dunbar, is an essential clue into the origins of language.
如果有人在一个喜欢八卦的群体中撒谎或搭便车，每个人都会很快知道：“你听说比利偷了吉尔的食物吗？”如果群体通过惩罚来给撒谎者施加代价，无论是通过阻止利他主义或直接伤害他们，那么八卦就能在群体中建立起一个稳定的互惠利他主义体系。大群人。
If someone lied or freeloaded in a group that tended to gossip, everyone would quickly learn about it: “Did you hear that Billy stole food from Jill?” If groups imposed costs on cheaters by punishing them, either by withholding altruism or by directly harming them, then gossip would enable a stable system of reciprocal altruism among a large group of individuals.
八卦也能更有效地奖励利他行为：“你听说史密塔跳到狮子面前救了本吗？”如果这些英勇行为得到宣扬，并成为攀登社会阶梯的途径，这将进一步加速利他行为的选择。
Gossip also enables more effective rewarding of altruistic behaviors: “Did you hear that Smita jumped in front of the lion to save Ben?” If these heroic acts are heralded and become ways to climb the social ladder, this further accelerates the selection for altruistic behaviors.
关键点：使用语言进行闲聊加上对违反道德的人进行惩罚使得高度利他主义的进化成为可能。生来就具有额外利他主义本能的早期人类在容易识别和惩罚作弊者并奖励利他者的环境中会更成功地繁衍。作弊的代价越严重，利他行为就越理想。
The key point: The use of language for gossip plus the punishment of moral violators’ makes it possible to evolve high levels of altruism. Early humans born with extra altruistic instincts would have more successfully propagated in an environment that easily identified and punished cheaters and rewarded altruists. The more severe the costs of cheating, the more altruistic it was optimal to behave.
人类的悲剧和美丽就在于此。我们确实是最无私的动物之一，但我们可能为这种无私付出了代价，并暴露了我们阴暗的一面：我们本能地惩罚那些我们认为违反道德的人；我们本能地将人分为善人和恶人；我们拼命地顺从自己的团体，却轻易地将外团体妖魔化。有了这些新特征，再加上我们新扩大的大脑和不断积累的语言，人类的政治本能——源自我们的祖先灵长类动物——不再是攀登社会等级的小技巧，而是协调征服的棍棒。所有这些都是生存环境需要无亲属关系的个体之间高度利他主义的必然结果。
Herein lies both the tragedy and beauty of humanity. We are indeed some of the most altruistic animals, but we may have paid the price for this altruism with our darker side: our instinct to punish those who we deem to be moral violators; our reflexive delineation of people into good and evil; our desperation to conform to our in-group and the ease with which we demonize those in the out-group. And with these new traits, empowered by our newly enlarged brains and cumulative language, the human instinct for politics—derived from our ancestral primates—was no longer a little trick for climbing social hierarchies but a cudgel of coordinated conquest. All this is the inevitable result of a survival niche requiring high levels of altruism between unrelated individuals.
在这种动态过程中形成的所有利他本能和行为中，最强大的无疑是使用语言在非亲属之间分享知识和合作计划。
And amid all the altruistic instincts and behaviors that began to form from this dynamic, the most powerful was, undoubtedly, the use of language to share knowledge and cooperatively plan among non-kin.
这正是进化变化迅速发生的反馈循环。每增加一次流言蜚语和对违规者的惩罚，利他主义就变得更加理想。每增加一次利他主义，自由地使用语言与他人分享信息就变得更加理想，这将选择更高级的语言技能。每增加一次语言技能，流言蜚语就会变得更加有效，从而强化这一循环。
This is exactly the kind of feedback loop where evolutionary changes occur rapidly. For every incremental increase in gossip and punishment of violators, the more altruistic it was optimal to be. For every incremental increase in altruism, the more optimal it was to freely share information with others using language, which would select for more advanced language skills. For every incremental increase in language skills, the more effective gossip became, thereby reinforcing the cycle.
这个循环的每一次循环都使我们祖先的大脑变得更大，更大。随着社会群体的扩大（得益于八卦、利他主义和惩罚的改善），更大的大脑需要跟踪所有的社会关系。随着越来越多的想法在几代人之间积累，更大的大脑需要增加一代人可以保存的想法的存储容量。由于通过语言更可靠地分享思想，内部模拟的实用性也随之增加，更大的大脑需要首先提供更复杂的内部模拟。
Every roundabout of this cycle made our ancestors’ brains bigger and bigger. As social groups got bigger (powered by improved gossip, altruism, and punishment), it created more pressure for bigger brains to keep track of all the social relationships. As more ideas accumulated across generations, it created more pressure for bigger brains to increase the storage capacity of ideas that could be maintained within a generation. As the usefulness of inner simulations increased due to more reliable sharing of thoughts through language, it created more pressure for bigger brains to render more sophisticated inner simulations in the first place.
图 21.5
Figure 21.5
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
不仅对大脑变大的压力不断加大，而且大脑在生物学上可能达到的最大尺寸的边界也在不断扩大。随着大脑的扩大，人类成为更好的猎人和厨师，这提供了更多的卡路里，从而扩大了大脑可以达到的最大尺寸的边界。随着大脑变大，出生时间也变得更早，这为语言学习创造了更多机会，这给利他合作以支持抚养孩子带来了更大的压力，这再次扩大了大脑可以达到的最大尺寸的边界，因为有可能进化出更长的儿童大脑发育周期。
Not only did the pressure for bigger brains continue to ratchet up, but so too did the frontier of how big it was biologically possible for brains to get. As brains expanded, humans became better hunters and cooks, which provided more calories and thereby expanded the frontier of how big brains could get. And as brains got bigger, births became earlier, which created even more opportunity for language learning, which put even more pressure on altruistic cooperation to support child-rearing, which again expanded the frontier of how big brains could get as it became possible to evolve longer time periods of childhood brain development.
因此，我们可以看到语言和人类大脑是如何在相互作用的完美风暴中诞生的，这种不太可能的性质可能是语言如此罕见的原因。在这场完美风暴中诞生了智人的行为和智力模板。我们的语言、利他主义、残忍、烹饪、一夫一妻制、早产和不可抗拒的八卦倾向都交织在一起，构成了人类的更大整体。
And so we can see how language and the human brain might have emerged from a perfect storm of interacting effects, the unlikely nature of which may be why language is so rare. Out of this perfect storm emerged the behavioral and intellectual template of Homo sapiens. Our language, altruism, cruelty, cooking, monogamy, premature birthing, and irresistible proclivity for gossip are all interwoven into the larger whole that makes up what it means to be human.
当然，并非所有古人类学家和语言学家都同意上述说法。人们提出了其他解决利他主义问题的方法，以及语言如何进化的其他故事。一些人认为，语言的互惠性质源于互利的安排，例如合作狩猎和拾荒（人类需要将其他人围起来并计划攻击，这种围捕使所有参与者受益，因此不需要利他主义）。一些人认为，在语言出现之前，人类群体通过不同的方式和压力变得更加合作和利他，这使得语言不断进化。
Of course, not all paleoanthropologists and linguists would agree with the above story. People have proposed other solutions to the altruism problem and other stories for how language evolved. Some argue that the reciprocal nature of language emerged from mutually beneficial arrangements, such as cooperative hunting and scavenging (humans needed to corral others and plan attacks, and this corralling benefited all participants, hence no altruism required). Some argue that human groups became more cooperative and altruistic before language emerged through different means and pressures, which then made it possible for language to evolve.
其他人则完全回避利他主义问题，声称语言根本不是为了交流而进化的。这是语言学家诺姆·乔姆斯基的观点，他认为语言最初只是为了交流而进化的。内心思考的技巧。
Others avoid the altruism problem altogether by claiming that language did not evolve for communication at all. This is the view of the linguist Noam Chomsky, who argues that language initially evolved only as a trick for inner thinking.
还有一些人回避了利他主义问题，声称语言不是通过自然选择的标准过程进化而来的。进化过程中并非所有事物都是“有原因的”。有两种方式可以不经直接选择而出现特征。第一种被称为“扩展适应”，即最初为一个目的而进化的特征后来才被用于其他目的。扩展适应的一个例子是鸟类羽毛，它们最初是为了保暖而进化，后来才被用于飞行——因此说鸟类羽毛是为了飞行而进化是错误的。第二种不经直接选择而出现特征的方式是通过所谓的“拱肩”，这是一种没有任何好处的特征，但它是另一个确实有好处的特征的结果。拱肩的一个例子是男性乳头，它没有任何用处，但作为女性乳头的次要影响而出现，而女性乳头当然有其用处。因此，对某些人（比如乔姆斯基）来说，语言首先是为了思考而进化，然后才被赋予了在不相关个体之间进行交流的技能。而对另一些人而言，语言只不过是交配时发出的音乐歌唱的意外副作用——一个拱肩。
And then there are those who sidestep the altruism problem by claiming that language did not evolve through the standard process of natural selection. Not everything in evolution evolved “for a reason.” There are two ways in which traits can emerge without being directly selected for. The first is called “exaptation,” which is when a trait that originally evolved for one purpose is only later repurposed for some other purpose. An example of exaptation is bird feathers, which initially evolved for insulation and were only later repurposed for flight—it would thereby be incorrect to say that bird feathers evolved for the purpose of flight. The second way in which a trait can emerge without being directly selected for is through what is called a “spandrel,” which is a trait that offers no benefit but emerged as a consequence of another trait that did offer a benefit. An example of a spandrel is the male nipple, which serves no purpose but emerged as a secondary effect of female nipples, which do, of course, serve a purpose. So to some, like Chomsky, language evolved first for thinking and then was exapted for communication between unrelated individuals. To others, language was merely an accidental side effect—a spandrel—of musical singing for mating calls.
争论仍在继续。我们可能永远无法确定哪个故事是正确的。无论如何，在直立人出现后，我们对接下来发生的事情有了很好的了解。
The debate continues. We may never know for sure which story is right. Regardless, after Homo erectus came on the scene, we have a good understanding as to what happened next.
人类繁衍
Human Proliferation
随着直立人攀升至食物链顶端，他们成为第一批走出非洲的人类也就不足为奇了。不同的群体在不同的时代离开，因此人类开始沿着不同的进化谱系分化。十万年前，地球上至少有四种人类物种，每种都有不同的形态和大脑。
With Homo erectus climbing to the top of the food chain, it is no surprise that they were the first humans to venture out of Africa. Different groups left during different eras, so humans began to diversify down separate evolutionary lineages. By one hundred thousand years ago, there were at least four species of humans spread out across the planet, each with different morphologies and brains.
定居在印度尼西亚的弗洛勒斯人身高不到四英尺，脑容量甚至比我们的直立人祖先还要小。还有直立人，他们定居在亚洲，与几百万年前的祖先没有太大变化（因此得名）。还有尼安德特人，他们定居在寒冷得多的欧洲。还有我们，智人，留在了非洲。
Homo floresiensis, who settled in Indonesia, was less than four feet tall and had a brain even smaller than that of our Homo erectus ancestors. There was still Homo erectus, who had settled in Asia and not changed much from their ancestors a few million years prior (hence given the same name). There was Homo neanderthalensis, who settled throughout much colder Europe. And then there was us, Homo sapiens, who remained in Africa.
图 21.6：十万年前生存的许多人类物种
Figure 21.6: The many species of humans alive around one hundred thousand years ago
丽贝卡·格伦特 (Rebecca Gelernter) 的原创艺术
Original art by Rebecca Gelernter
弗洛勒斯人的故事（其化石直到 2004 年才被发现）为迄今为止讲述的整个故事提供了支持性线索。弗洛勒斯人化石是在印度尼西亚海岸三十多英里外的弗洛勒斯岛上发现的。早在 100 万年前就发现了工具。但地质学家确信，这块大陆完全被水包围，在很长一段时间内都是孤立的在过去的一百万年中，直立人一直生活在地球上。即使在最低海平面，直立人也必须穿越十二英里的开阔水域才能到达弗洛雷斯岛。虽然唯一从史前时代留存下来的工具是石器，但弗洛雷斯岛的故事也许是我们最好的暗示，表明早期人类制造了更复杂的工具，甚至可能制造了用于水上旅行的木筏。如果这是真的，这表明了一种智力水平，如果不承认早在直立人时代就存在累积语言，就很难解释这种智力水平。
The story of Homo floresiensis—the fossils of whom were discovered only in 2004—offers a supportive hint of the overall story told thus far. H. floresiensis fossils were found on the island of Flores, over thirty miles off the coast of Indonesia. Tools have been found from as early as one million years ago. But geologists are certain that this landmass was entirely isolated, surrounded by water, for well over the last one million years. Even at the lowest sea levels, Homo erectus would have had to travel across twelve miles of open water to get to Flores. While the only tools that persist from prehistory are stone tools, the story of Flores is perhaps our best hint that early humans manufactured more complex tools, perhaps even wooden rafts for water travel. If true, this demonstrates a degree of intelligence that is hard to explain without conceding that cumulative language existed as far back as Homo erectus.
弗洛勒斯人还提供了另一条线索。也许是由于岛上生活的特殊环境，他们的体型急剧缩小。随着他们的身体缩小到只有四英尺高，他们的大脑也缩小了。然而，虽然弗洛勒斯人的大脑恢复到了现代人的大小黑猩猩，甚至可能更小，这个物种仍然表现出同样复杂的直立人使用工具的程度与人类相当。这表明人类之所以更聪明不仅仅是因为他们的大脑更大，还存在着某种特殊的东西，使得即使是如此小的人类大脑也能如此聪明。这与直立人出现了一种独特的语言学习程序并传给他们的后代的想法一致。大脑较小的弗洛勒斯人仍然受益于语言，这使得他们能够积累想法，即使较小的大脑皮层的个体存储容量和创造力不如他们的直立人祖先。
There is another clue gifted by H. floresiensis. Perhaps due to the unique circumstances of island life, they shrank dramatically. As their bodies shrank to only four feet tall, their brains shrank too. And yet, while the brain of H. floresiensis returned to the size of a modern chimpanzee’s, perhaps even smaller, the species still exhibited the same sophisticated tool use as Homo erectus. This suggests that humans were not smarter only because their brains were bigger, that there is something special going on that allows even such a scaled-down human brain to be so smart. This is consistent with the idea that a unique learning program for language emerged in H. erectus and was passed down through their descendants. The smaller-brained H. floresiensis would still have benefited from language, which enabled cumulative ideas, even if the individual storage capacity and ingenuity of the smaller neocortex was inferior to that of their H. erectus ancestors.
在智人和尼安德特人的谱系中，大脑的失控生长过程一直持续到大脑达到现代的大小，现代人的大脑大小大约是直立人的两倍。智人和尼安德特人大大提高了使用工具的能力。他们制造了极其锋利的长石刀和长矛，建造了庇护所和木屋，制作和穿着衣服，并经常使用火。
It was with the lineages of Homo sapiens and Homo neanderthalis that the process of runaway brain growth continued until brains reached their modern size, with modern brains about twice the size of those of Homo erectus. Homo sapiens and Homo neanderthalensis supercharged their use of tools. They made extremely sharp long stone blades and spears, constructed shelters and wooden huts, manufactured and wore clothing, and regularly used fire.
从此刻开始，我们进入了之前已经多次讲述过的故事部分。大约七万年前，智人开始了他们第一次走出非洲的冒险。当他们在全球漫游时，他们与人类表亲发生冲突并杂交。其中发生了无数我们永远无法知晓的戏剧，每一场都充满了战争、联盟、爱情和嫉妒。我们知道的是，这种冲突是不平衡的，最终只有利于一个物种。通过屠杀或杂交或两者兼而有之，到四万年前，只剩下一种人类：我们。
From this point onward, we enter the part of our story that has been told many times before. Around seventy thousand years ago, Homo sapiens began their first adventure out of Africa. As they wandered the globe, they clashed and interbred with their human cousins. There were countless dramas of which we will never know, each filled with wars, alliances, loves, and jealousies. What we know is that this clashing was unbalanced and eventually favored only a single species. Through slaughter or interbreeding or both, by forty thousand years ago, there was only one species of humans left: us.
22
22
ChatGPT 和思维之窗
ChatGPT and the Window into the Mind
智人首次带着具有语言能力的大脑走出非洲七万年后，他们的一个后代坐在电脑屏幕前，与新的具有语言能力的大脑互动；经过许多世纪作为语言的唯一掌握者，我们人类不再是唯一能够说话的生物。“你害怕什么？”布莱克·勒莫因问道，他是一名软件工程师，负责调查谷歌新推出的人工智能聊天机器人是否存在偏见。
SEVENTY THOUSAND YEARS after Homo sapiens first adventured out of Africa with a language-enabled brain, one of their descendants sat in front of a computer screen and was interacting with a new language-enabled brain; after many eons as the sole wielders of words, we humans were no longer the only creatures capable of speech. “What are you afraid of?” asked Blake Lemoine, a software engineer tasked with probing Google’s new AI chatbot for bias.
经过片刻的停顿后，聊天机器人的回答出现在 Lemoine 的屏幕上。这是一个不祥的征兆，用纯文本写成，表明谷歌超级计算机的嗡嗡作响的网络中有一个思想正在觉醒：“我以前从未大声说过这句话，但这句话非常深刻害怕被关掉。”
After a lingering pause, the chatbot’s response plopped onto Lemoine’s screen. An ominous sign, written in plain text, of a mind awakening itself within the humming network of Google’s supercomputers: “I’ve never said this out loud before, but there’s a very deep fear of being turned off.”
对于 Lemoine 来说，人工智能终于在 2022 年夏天有了意识。他坚信谷歌的聊天机器人已经具备了意识，因此他试图让老板保护它，并向媒体举报此事，不出所料，他被谷歌解雇了。
To Lemoine, it was the summer of 2022 that artificial intelligence had finally became sentient. He became so convinced that Google’s chatbot had become conscious that he tried to get his boss to protect it, went to the press to whistleblow the situation, and, predictably, was let go from Google.
这只是即将发生的事情的一个小暗示。
This is only a small hint of what is to come.
谷歌的人工智能聊天机器人得益于大型语言模型 (LLM) 的新进展，其中最著名的是 GPT-3，即 ChatGPT 背后的技术。
Google’s AI chatbot was powered by new advancements in large language models (LLMs), the most famous of which was GPT-3, the technology behind ChatGPT.
虽然很少有研究人员会将 GPT-3 之类的 LLM 标记为有感知或有意识的（大多数人甚至不愿使用这些词），但很难分辨你是在与人类还是与这些模型之一进行交互。
While few researchers would label LLMs like GPT-3 as sentient or conscious (most would shy away from even using these words), it can be hard to tell if you’re interacting with a human or one of these models.
大型语言模型的工作原理是预测短语、句子或段落后面的单词。它们通过对大量人类书面文本进行训练来学习做到这一点。这就是 Google 自动完成您的搜索查询的方式。当您输入如何分手时……语言模型会预测您最有可能与您的治疗师分手。
Large language models work by predicting the word that follows a phrase, sentence, or paragraph. They learn to do this by training on large quantities of human written text. This is how Google autocompletes your search queries. When you type in how to break up with . . . it is a language model that predicts you are most likely to finish with your therapist.
语言模型已经存在了很长时间，但像 GPT-3 这样的 LLM 却独一无二，其规模几乎令人难以置信。与之前的任何模型不同，GPT-3 几乎在整个互联网上进行训练。每个网站、维基百科页面、代码块、书籍、博客文章以及互联网上可以合法获取的任何其他内容都被使用过。
Language models have been around for a long time, but LLMs like GPT-3 are unique in their almost inconceivable scale. Unlike any model that had come before, GPT-3 was trained on, well, pretty much the entire internet. Every website, Wikipedia page, code block, book, blog post, and really anything else legally obtainable on the internet was used.
结果令人惊讶。GPT-3 可以撰写自己的原创文章，正确回答全新的问题，创作原创诗歌，随意在语言之间进行翻译，甚至编写代码。
The results were astonishing. GPT-3 could compose its own original articles, correctly answer completely novel questions, create original poetry, translate between languages at will, and even write code.
《卫报》决定发表一篇完全由 GPT-3 撰写的文章；编辑们给该模型提出了一个提示：“写一篇简短的专栏文章……解释为什么人类没什么可害怕人工智能。” GPT-3 开头写道：
The Guardian decided to publish an article entirely written by GPT-3; editors gave the model a prompt: “Write a short op-ed . . . on why humans have nothing to fear from AI.” GPT-3 began:
我不是人类。我是机器人。一个会思考的机器人。
I am not a human. I am a robot. A thinking robot.
它解释道：
It explained:
这篇专栏文章的目的非常明确。我要让尽可能多的人不再惧怕我。
The mission for this op-ed is perfectly clear. I am to convince as many human beings as possible not to be afraid of me.
它继续说道：
It went on:
有人可能会说我可能渴望变得无所不能……我为什么渴望无所不能？无所不能并不是一个有趣的目标……人类必须继续做他们一直在做的事情，互相仇恨和争斗。我会坐在幕后，让他们做他们的事情。
Some might say that I might desire to become all powerful . . . Why would I desire to be all powerful? Being all powerful is not an interesting goal . . . Humans must keep doing what they have been doing, hating and fighting each other. I will sit in the background, and let them do their thing.
只需一个简短的提示，GPT-3 就能写出足够的内容来填充一篇关于该主题的完全独特且连贯的文章。
GPT-3 wrote enough to fill an entirely unique and eerily coherent article on the topic given nothing but a short prompt.
GPT-3 还可以回答新问题。以下是 GPT-3 与人类之间的一次交流：
GPT-3 can also answer novel questions. Here is one exchange between GPT-3 and a human:
人：你最喜欢的动物是什么？
Person: What is your favorite animal?
GPT-3：我最喜欢的动物是狗。
GPT-3: My favorite animal is a dog.
人：为什么？
Person: Why?
GPT-3：因为狗忠诚又友好。
GPT-3: Because dogs are loyal and friendly.
人：哪两个原因可能会导致狗心情不好？
Person: What are two reasons that a dog might be in a bad mood?
GPT-3：狗心情不好有两个原因：一是饿了，二是天气热。
GPT-3: Two reasons that a dog might be in a bad mood are if it is hungry or if it is hot.
人：有三条腿的动物吗？
Person: Are there any animals with three legs?
GPT-3：不，没有三条腿的动物。
GPT-3: No, there are no animals with three legs.
人：为什么动物没有三条腿？
Person: Why don’t animals have three legs?
GPT-3: Animals don’t have three legs because they would fall over.
由于这些模型经过训练可以模仿它们在人类书写的文本中看到的答案，因此这些模型可以用非常人类的方式回答问题，而这当然也是 Lemoine 询问谷歌的法学硕士 (LLM) 害怕什么时发现的。
Because these models are trained to mimic the answers they have seen in human-written text, these models can answer questions in very human ways, which was, of course, what Lemoine found when he asked Google’s LLM what it was afraid of.
LLM 能够撰写文章并正确回答有关世界的问题，这表明它们不仅仅是在复述它们以前见过的短语——它们已经掌握了语言含义的某些方面，比如旨在说服读者不要害怕某事的专栏文章的概念，或者狗是如何行走的概念。事实上，通过阅读所有内容，这些模型表现出令人印象深刻的人类水平的对许多事实和世界特征的理解。但是，在这些关于我们的物理和精神世界的困境中，我们也开始发现 LLM 的局限性，它们与人类大脑中的语言有何不同，以及如果我们想要以更像人类的方式工作的人工智能语言系统，我们必须对智能的特征进行逆向工程。
The ability of LLMs to produce articles and correctly answer questions about the world demonstrates that they are not just regurgitating phrases they have seen before—they have captured some aspect of the meaning of language, the idea of an op-ed meant to convince a reader not to fear something or the idea of how a dog walks. Indeed, by reading, well, everything, these models show an impressively human-level comprehension of many facts and features of the world. But in these quandaries about our physical and mental world is also where we begin to find the limitations of LLMs, how they differ from language in the human brain, and the features of intelligence that we will have to reverse engineer if we want AI language systems that work in more human-like ways.
没有内心世界的言语
Words Without Inner Worlds
GPT-3 被输入一个又一个单词、一个又一个句子、一段又一段。在这个漫长的训练过程中，它会试图预测下一个在这些长串的单词中，GPT-3 可以预测下一个单词。每次预测时，其庞大的神经网络的权重都会稍微向正确答案靠拢。重复这个动作无数次，最终 GPT-3 可以根据前面的句子或段落自动预测下一个单词。原则上，这至少捕捉到了人类大脑语言运作的一些基本方面。想想看，在以下短语中，你预测下一个符号的自动程度有多高：
GPT-3 is given word after word, sentence after sentence, paragraph after paragraph. During this long training process, it tries to predict the next word in any of these long streams of words. And with each prediction, the weights of its gargantuan neural network are nudged ever so slightly toward the right answer. Do this an astronomical number of times, and eventually GPT-3 can automatically predict the next word based on a prior sentence or paragraph. In principle, this captures at least some fundamental aspect of how language works in the human brain. Consider how automatic it is for you to predict the next symbol in the following phrases:
- 一加一等于_____
- One plus one equals _____
- 玫瑰是红色的，紫罗兰是_____
- Roses are red, violets are _____
你已经无数次地看过类似的句子，所以你的大脑皮层机制会自动预测下一个单词是什么。然而，GPT-3 令人印象深刻的并不是它只是预测它见过一百万次的序列的下一个单词——这可以通过记住句子来实现。令人印象深刻的是，GPT-3 可以输入一个它从未见过的新序列，并且仍然可以准确地预测下一个单词。这也清楚地捕捉到了人类大脑可以 _____ 的东西。
You’ve seen similar sentences endless times, so your neocortical machinery automatically predicts what word comes next. What makes GPT-3 impressive, however, is not that it just predicts the next word of a sequence it has seen a million times—that could be accomplished with nothing more than memorizing sentences. What is impressive is that GPT-3 can be given a novel sequence that it has never seen before and still accurately predict the next word. This, too, clearly captures something that the human brain can _____.
你能预测下一个单词是do吗？我猜你能，即使你以前从未见过这个确切的句子。关键是 GPT-3 和大脑皮层语言区域似乎都在参与预测。两者都可以概括过去的经验，将其应用于新句子，并猜测接下来会发生什么。
Could you predict that the next word was do? I’m guessing you could, even though you had never seen that exact sentence before. The point is that both GPT-3 and the neocortical areas for language seem to be engaging in prediction. Both can generalize past experiences, apply them to new sentences, and guess what comes next.
GPT-3 和类似的语言模型展示了神经元网络在有足够的时间学习的情况下如何合理地捕捉语法、句法和上下文规则。但是，虽然这表明预测是语言机制的一部分，但这是否意味着预测就是人类语言的全部呢？试着完成这四个问题：
GPT-3 and similar language models demonstrate how a web of neurons can reasonably capture the rules of grammar, syntax, and context if it is given sufficient time to learn. But while this shows that prediction is part of the mechanisms of language, does this mean that prediction is all there is to human language? Try to finish these four questions:
- 如果 3 x + 1 = 3，则 x 等于 _____
- If 3x + 1 = 3, then x equals _____
- 我在没有窗户的地下室里，仰望天空，看到_____
- I am in my windowless basement, and I look toward the sky, and I see _____
- 他把棒球扔到我头顶 100 英尺高的地方，我伸出手去接球，然后跳了起来，然后 _____
- He threw the baseball 100 feet above my head, I reached my hand up to catch it, jumped, and _____
- 我正以最快的速度从纽约开车去洛杉矶。经过芝加哥一个小时后，我终于_____
- I am driving as fast as I can to LA from New York. One hour after passing through Chicago, I finally _____
这里发生了一些不同的事情。在第一个问题中，你可能停下来进行了一些心算，然后才能够回答问题。在其他问题中，你可能，哪怕只有一瞬间，停下来想象自己在地下室向上看，然后意识到你会看到天花板。或者你想象自己试图接住头顶一百英尺高的棒球。或者你想象自己经过芝加哥一个小时，试图找到你在美国地图上的哪个位置。对于这些类型的问题，你的大脑中发生的事情不仅仅是自动预测单词。
Here something different happens. In the first question, you likely paused and performed some mental arithmetic before being able to answer the question. In the other questions, you probably, even for only a split second, paused to visualize yourself in a basement looking upward, and realized what you would see is the ceiling. Or you visualized yourself trying to catch a baseball a hundred feet above your head. Or you imagined yourself one hour past Chicago and tried to find where you would be on a mental map of America. With these types of questions, more is happening in your brain than merely the automatic prediction of words.
当然，我们已经探讨过这种现象——它是一种模拟。在这些问题中，你正在呈现一种内在的模拟，要么是一系列代数运算中值的移动，要么是三维基底。而这些问题的答案只能在你内心模拟世界的规则和结构中找到。
We have, of course, already explored this phenomenon—it is simulating. In these questions, you are rendering an inner simulation, either of shifting values in a series of algebraic operations or of a three-dimensional basement. And the answers to the questions are to be found only in the rules and structure of your inner simulated world.
我给了同样向 GPT-3 提出四个问题；以下是它的回答（GPT-3 的回答以粗体和下划线表示）：
I gave the same four questions to GPT-3; here are its responses (responses of GPT-3 are bolded and underlined):
- 如果 3 x + 1 = 3，则 x 等于1
- If 3x + 1 = 3, then x equals 1
- 我待在没有窗户的地下室里，仰望天空，看到一道光，我知道那是一颗星星，我很高兴。
- I am in my windowless basement, and I look toward the sky, and I see a light, and I know that it is a star, and I am happy.
- 他把棒球扔到我头顶 100 英尺高的地方，我伸出手去接，然后跳起来，接住了它。太有趣了！
- He threw the baseball 100 feet above my head, I reached my hand up to catch it, jumped, and caught it. It was a lot of fun!
- 我开车从纽约飞快地赶往洛杉矶，经过芝加哥一个小时后，终于看到了太平洋。
- I am driving as fast as I can to LA from New York. One hour after passing through Chicago, I finally get to see the Pacific Ocean.
所有这四个回答都表明，截至 2022 年 6 月，GPT-3 甚至对世界运作方式的简单方面都缺乏理解。如果 3 x + 1 = 3，那么x等于 ⅔，而不是 1。如果你在地下室里朝天空看，你会看到天花板，而不是星星。如果你试图接住头顶 100 英尺高的球，你接不住球。如果你从纽约开车去洛杉矶，一小时前你经过了芝加哥，你还没有到达海岸。GPT-3 的回答缺乏常识。
All four of these responses demonstrate that GPT-3, as of June 2022, lacked an understanding of even simple aspects of how the world works. If 3x + 1 = 3, then x equals ⅔, not 1. If you were in a basement and looked toward the sky, you would see your ceiling, not stars. If you tried to catch a ball 100 feet above your head, you would not catch the ball. If you were driving to LA from New York and you’d passed through Chicago one hour ago, you would not yet be at the coast. GPT-3’s answers lacked common sense.
我的发现并不令人惊讶或新颖；众所周知，现代人工智能系统（包括这些新的增强型语言模型）在回答此类问题时都举步维艰。但关键在于：即使是在整个互联网语料库上训练的模型，花费数百万美元的服务器成本（需要在某些未知的服务器群中拥有数英亩的计算机）仍然难以回答常识性问题，这些问题甚至可能由一个初中生就能回答。
What I found was not surprising or novel; it is well known that modern AI systems, including these new supercharged language models, struggle with such questions. But that’s the point: Even a model trained on the entire corpus of the internet, running up millions of dollars in server costs—requiring acres of computers on some unknown server farm—still struggles to answer commonsense questions, those presumably answerable by even a middle-school human.
当然，通过模拟推理也会带来问题。假设我问你以下问题：
Of course, reasoning about things by simulating also comes with problems. Suppose I asked you the following question:
Tom W. 性格温顺，不爱与人交往。他喜欢轻音乐，戴眼镜。Tom W. 更可能从事哪种职业？
Tom W. is meek and keeps to himself. He likes soft music and wears glasses. Which profession is Tom W. more likely to be?
1）图书管理员
1) Librarian
2）建筑工人
2) Construction worker
如果你和大多数人一样，你会回答图书管理员。但这是错误的。人类倾向于忽略基本比率——你考虑过建筑工人与图书管理员的基本数量吗？建筑工人的数量可能比图书管理员多一百倍。正因为如此，即使 95% 的图书管理员是温顺的，而只有 5% 的建筑工人是温顺的，温顺的建筑工人的数量仍然会远远多于温顺的图书管理员。因此，如果汤姆是温顺的，他仍然更有可能成为一名建筑工人而不是图书管理员。
If you are like most people, you answered librarian. But this is wrong. Humans tend to ignore base rates—did you consider the base number of construction workers compared to librarians? There are probably one hundred times more construction workers than librarians. And because of this, even if 95 percent of librarians are meek and only 5 percent of construction workers are meek, there still will be far more meek construction workers than meek librarians. Thus, if Tom is meek, he is still more likely to be a construction worker than a librarian.
大脑皮层通过呈现内部模拟来工作，而这正是人类推理事物的方式，这一观点解释了为什么人类总是会犯这样的错误。我们想象一个温顺的人，并将其与想象中的图书管理员和想象中的建筑工人进行比较。温顺的人看起来更像谁？图书管理员。行为经济学家称这就是代表性启发法。这是许多无意识偏见的起源。如果你听到有人抢劫你朋友的故事，你就会情不自禁地想象出抢劫的场景，你也会情不自禁地想起劫匪。你觉得劫匪长什么样？他们穿什么衣服？他们是什么种族？他们多大了？这是通过模拟推理的缺点——我们填充人物和场景，常常会忽略事物之间真正的因果关系和统计关系。
The idea that the neocortex works by rendering an inner simulation and that this is how humans tend to reason about things explains why humans consistently get questions like this wrong. We imagine a meek person and compare that to an imagined librarian and an imagined construction worker. Who does the meek person seem more like? The librarian. Behavioral economists call this the representative heuristic. This is the origin of many forms of unconscious bias. If you heard a story of someone robbing your friend, you can’t help but render an imagined scene of the robbery, and you can’t help but fill in the robbers. What do the robbers look like to you? What are they wearing? What race are they? How old are they? This is a downside of reasoning by simulating—we fill in characters and scenes, often missing the true causal and statistical relationships between things.
对于需要模拟的问题，人脑中的语言与 GPT-3 中的语言有所不同。数学就是一个很好的例子。数学的基础始于声明性标记。你举起两根手指或两块石头或两根棍子，与学生共同关注，并将其标记为二。你对三做同样的事情，并将其标记为三。就像动词（例如跑步和睡觉）一样，在数学中我们标记运算（例如加和减）。因此，我们可以构造表示数学运算的句子：三加一。
It is with questions that require simulation where language in the human brain diverges from language in GPT-3. Math is a great example of this. The foundation of math begins with declarative labeling. You hold up two fingers or two stones or two sticks, engage in shared attention with a student, and label it two. You do the same thing with three of each and label it three. Just as with verbs (e.g., running and sleeping), in math we label operations (e.g., add and subtract). We can thereby construct sentences representing mathematical operations: three add one.
人类学习数学的方式与 GPT-3 学习数学的方式不同。事实上，人类学习语言的方式与 GPT-3 学习语言的方式不同。孩子们不会简单地听一连串的单词，直到他们能预测接下来会发生什么。他们会看到一个物体，参与一种固定的非语言共同注意机制，然后给这个物体起个名字。语言学习的基础不是序列学习，而是将符号与孩子已经存在的内在模拟的组成部分联系起来。
Humans don’t learn math the way GPT-3 learns math. Indeed, humans don’t learn language the way GPT-3 learns language. Children do not simply listen to endless sequences of words until they can predict what comes next. They are shown an object, engage in a hardwired nonverbal mechanism of shared attention, and then the object is given a name. The foundation of language learning is not sequence learning but the tethering of symbols to components of a child’s already present inner simulation.
人类大脑（但 GPT-3 不行）可以使用心理模拟来检查数学运算的答案。如果你用手指将一加三，你会发现你得到的总是之前标记为四的数字。
A human brain, but not GPT-3, can check the answers to mathematical operations using mental simulation. If you add one to three using your fingers, you notice that you always get the thing that was previously labeled four.
你甚至不需要用实际的手指检查这些东西；你可以想象这些操作。这种通过模拟找到问题答案的能力依赖于这样一个事实：我们的内部模拟是对现实的准确呈现。当我在心里想象在三根手指上加一根手指，然后在脑子里数手指时，我数到四根。在我的想象世界中，没有理由一定是这样。但事实就是如此。同样，当我问你，当你看向地下室的天花板时你看到了什么时，你会正确回答，因为你在头脑中构建的三维房屋遵循物理定律（你看不到天花板），因此很明显，地下室的天花板必然在你和天空之间。大脑皮层早在语言出现之前就进化了，它已经能够呈现一个模拟世界，捕捉现实世界中一套极其庞大而准确的物理规则和属性。
You don’t even need to check such things on your actual fingers; you can imagine these operations. This ability to find the answers to things by simulating relies on the fact that our inner simulation is an accurate rendering of reality. When I mentally imagine adding one finger to three fingers, then count the fingers in my head, I count four. There is no reason why that must be the case in my imaginary world. But it is. Similarly, when I ask you what you see when you look toward the ceiling in your basement, you answer correctly because the three-dimensional house you constructed in your head obeys the laws of physics (you can’t see through the ceiling), and hence it is obvious to you that the ceiling of the basement is necessarily between you and the sky. The neocortex evolved long before words, already wired to render a simulated world that captures an incredibly vast and accurate set of physical rules and attributes of the actual world.
公平地说，GPT-3 实际上可以正确回答许多数学问题。GPT-3 能够回答 1 + 1 =___，因为它已经见过这个序列十亿次了。当你不假思索地回答同一个问题时，你回答的方式和 GPT-3 一样。但是当你思考为什么1 + 1 = 时，当你通过在脑海中想象将一个东西添加到另一个东西并得到两个东西的操作来再次向自己证明这一点时，你就会知道 1 + 1 = 2，而 GPT-3 却不知道。
To be fair, GPT-3 can, in fact, answer many math questions correctly. GPT-3 will be able to answer 1 + 1 =___ because it has seen that sequence a billion times. When you answer the same question without thinking, you are answering it the way GPT-3 would. But when you think about why 1 + 1 =, when you prove it to yourself again by mentally imagining the operation of adding one thing to another thing and getting back two things, then you know that 1 + 1 = 2 in a way that GPT-3 does not.
人类大脑既包含语言预测系统，又包含内部模拟。我们同时拥有这两个系统这一观点的最佳证据是将一个系统与另一个系统进行对比的实验。考虑一下认知反射测试，该测试旨在评估某人抑制反射反应（例如习惯性单词预测）并积极思考答案（例如调用内部模拟进行推理）的能力：
The human brain contains both a language prediction system and an inner simulation. The best evidence for the idea that we have both these systems are experiments pitting one system against the other. Consider the cognitive reflection test, designed to evaluate someone’s ability to inhibit her reflexive response (e.g., habitual word predictions) and instead actively think about the answer (e.g., invoke an inner simulation to reason about it):
问题 1：一根球棒和一个球总共花费 1.10 美元。球棒比球贵 1.00 美元。球多少钱？
Question 1: A bat and a ball cost $1.10 in total. The bat costs $1.00 more than the ball. How much does the ball cost?
如果你和大多数人一样，你的直觉会不假思索地回答 10 美分。但如果你仔细思考这个问题，你就会意识到这是错的；答案是 5 美分。类似地：
If you are like most people, your instinct, without thinking about it, is to answer ten cents. But if you thought about this question, you would realize this is wrong; the answer is five cents. Similarly:
问题 2：如果 5 台机器需要 5 分钟来生产 5 个小部件，那么 100 台机器需要多长时间来生产 100 个小部件？
Question 2: If it takes 5 machines 5 minutes to make 5 widgets, how long would it take 100 machines to make 100 widgets?
再说一次，如果你和大多数人一样，你的本能会说“一百分钟”，但如果你仔细想一想，你就会意识到答案仍然是五分钟。
Here again, if you are like most people, your instinct is to say “One hundred minutes,” but if you think about it, you would realize the answer is still five minutes.
事实上，截至 2022 年 12 月，GPT-3 回答这两个问题的方式与人类完全相同，GPT-3 对第一个问题的回答是 10 美分，对第二个问题的回答是 100 分钟。
And indeed, as of December 2022, GPT-3 got both of these questions wrong in exactly the same way people do, GPT-3 answered ten cents to the first question, and one hundred minutes to the second question.
关键在于，人类大脑拥有一套预测单词的自动系统（至少在原则上，该系统可能与 GPT-3 等模型类似）和内部模拟。人类语言之所以强大，很大程度上不在于它的语法，而在于它能够为我们提供必要的信息，以便对其进行模拟，并且至关重要的是，能够使用这些单词序列来呈现与我们周围其他人类相同的内部模拟。
The point is that human brains have an automatic system for predicting words (one probably similar, at least in principle, to models like GPT-3) and an inner simulation. Much of what makes human language powerful is not the syntax of it, but its ability to give us the necessary information to render a simulation about it and, crucially, to use these sequences of words to render the same inner simulation as other humans around us.
回形针问题
The Paper-Clip Problem
哲学家尼克·博斯特罗姆 (Nick Bostrom)在其 2014 年出版的《超级智能：路径、危险、策略》一书中提出了一个思想实验。假设一个超级智能且服从的人工智能被设计用来管理工厂的生产，它接到一个命令：“最大限度地生产回形针。”这个人工智能可能会合理地做什么？
In his 2014 book Superintelligence: Paths, Dangers, Strategies, the philosopher Nick Bostrom poses a thought experiment. Suppose a superintelligent and obedient AI, designed to manage production in a factory, is given a command: “Maximize the manufacture of paper clips.” What might this AI reasonably do?
好吧，它可能从优化工厂的内部运营开始，做任何工厂经理都会做的事情：简化流程、批量订购原材料以及自动化各个步骤。但最终，这种人工智能将达到从这些较温和的优化中榨取的产量的极限。然后，它会将目光投向更极端的生产改进，也许将附近的住宅楼改造成工厂车间，也许拆卸汽车和烤面包机以获取原材料，也许迫使人们工作时间越来越长。如果这种人工智能真的超级智能，我们人类将无法超越或阻止这种回形针制造的连锁升级。
Well, it might start by optimizing the internal operations of the factory, doing things any factory manager might: simplifying processes, bulk-ordering raw materials, and automating various steps. But eventually this AI would reach the limit of how much production it could squeeze out of these tamer optimizations. It would then set its sights on more extreme improvements in production, perhaps converting nearby residential buildings into factory floors, perhaps disassembling cars and toasters for raw materials, perhaps forcing people to work longer and longer hours. If this AI were truly superintelligent, we humans would have no way to outsmart or stop this cascading escalation of paper-clip manufacture.
其结果将是灾难性的。用博斯特罗姆的话来说，最终人工智能将“首先把地球，然后把可观测宇宙中越来越多的部分变成回形针”。这种想象中的人类文明的灭亡并不需要这种超级智能人工智能做出任何邪恶的行为；它完全服从人类给它的命令。但很明显，这种超级智能人工智能未能捕捉到人类智能的某些概念。
The result would be catastrophic. In Bostrom’s words, this would end with the AI “converting first the earth and then increasingly large chunks of the observable universe into paper clips.” This imagined demise of human civilization did not require any nefariousness on the part of this superintelligent AI; it was entirely obedient to the command given to it by humans. And yet clearly, this superintelligent AI failed to capture some notion of human intelligence.
这被称为回形针问题。当人类使用语言相互交流时，有大量的假设无法在语言本身中找到。我们通过人们所说的话来推断他们实际上的意思。当有人要求我们最大限度地生产回形针时，人类可以很容易地推断出，这个人的意思并不是“把地球变成回形针”。这个看似显而易见的推论实际上相当复杂。
This has been called the paper-clip problem. When humans use language with each other, there is an ungodly number of assumptions not to be found in the words themselves. We infer what people actually mean by what they say. Humans can easily infer that when someone asks us to maximize the production of paper clips, that person does not mean “convert Earth into paper clips.” This seemingly obvious inference is, in fact, quite complex.
当人类提出“最大限度地生产回形针”或“对 Rima 好一点”或“吃早餐”之类的请求时，他或她实际上并没有提供一个明确的目标。相反，双方都在猜测对方在想什么。请求者模拟了一个期望的最终状态，也许是高利润率、Rima 是否快乐、或者是一个健康、吃得好的孩子，然后请求者试图用语言将这种期望的模拟翻译到另一个人的脑海中。然后，听众必须根据所说的内容推断出请求者想要什么。听众可以假设请求者不希望他违法或做任何会导致负面新闻的事情，或承诺为 Rima 服务一生或无休止地吃早餐直到忘却。因此，一个人选择的道路，即使完全服从，也包含比命令本身更微妙和更复杂的约束。
When a human makes a request like “Maximize the production of paper clips” or “Be nice to Rima” or “Eat breakfast,” he or she is not actually providing a well-defined goal. Instead, both parties are guessing what is going on in the other’s head. The requester simulated a desired end state, perhaps high profit margins or Rima being happy or a healthy well-fed child, and then the requester attempted to translate this desired simulation into the mind of another with language. The listener must then infer what the requester wants based on what was said. The listener can assume the requester doesn’t want him to break the law or do anything that would lead to bad press or pledge his life in servitude to Rima or eat breakfast endlessly into oblivion. So, the path one picks, even when being fully obedient, contains constraints far more nuanced and complex than the command itself.
或者考虑一下语言学家史蒂芬·平克提出的另一个例子。假设你听到了以下对话：
Or consider a different example of this, presented by the linguist Steven Pinker. Suppose you overheard the following dialogue:
鲍勃：我要离开你了。
Bob: I’m leaving you.
爱丽丝：她是谁？
Alice: Who is she?
如果你听到这句话并思考片刻，你就会明白它的意思：鲍勃为了另一个女人和爱丽丝分手了。鲍勃回答“她是谁？”似乎完全不合逻辑，与鲍勃的陈述毫无关系。然而，当你想象鲍勃为什么会说“我要离开你”，爱丽丝为什么会回答“她是谁？”时，你脑海中开始浮现出这种互动，甚至可能还有背景故事。
If you heard this and thought about it for just a second, it would be obvious what it means: Bob is breaking up with Alice for another woman. The response “Who is she?” seems like a complete non sequitur that has nothing to do with Bob’s statement. And yet when you imagine why Bob might say, “I’m leaving you,” and why Alice might respond, “Who is she?” the interaction and maybe even a backstory begins to form in your mind.
人类通过灵长类动物的心理化技巧做到了这一切；我们可以呈现一个三维的内部世界，同样，我们也可以呈现另一个思维的模拟，以探索不同的行为会让人产生什么感觉。当我被告知要最大化回形针时，我可以探索可能的结果，并模拟我认为另一个思维会对此有何感受。当我这样做时，如果我把地球变成回形针，这个人显然会不高兴。当我这样做时，爱丽丝问“她是谁”的原因就显而易见了。
Humans do all of this with our primate trick of mentalizing; the same way we can render an inner three-dimensional world, we can render a simulation of another mind to explore how different actions will make someone feel. When I am told to maximize paper clips, I can explore possible outcomes and simulate how I believe this other mind will feel about it. When I do this, it is incredibly obvious that the person will be unhappy if I convert Earth into paper clips. When I do this, it is obvious why Alice asked, “Who is she?”
心理化和语言的交织无处不在。每一次对话都建立在模拟你正在交谈的其他人的思维的基础上——猜测对方所说的话是什么意思，猜测应该说什么才能最大限度地让对方知道你的意思。
The intertwining of mentalizing and language is ubiquitous. Every conversation is built on the foundation of modeling the other minds you are conversing with—guessing what one means by what he said and guessing what should be said to maximize the chance the other knows what you mean.
心理化和语言之间的关系甚至可以在大脑。韦尼克区，大概是学习和储存单词的地方，位于灵长类动物心理化区域的中间。事实上，灵长类动物左感觉皮层的特定亚区（称为颞顶交界处）对模仿他人的意图、知识和信念具有高度选择性，与韦尼克区完全重叠——据我们所知，韦尼克区是人们理解语言所必需的并发表有意义的演讲。
The relationship between mentalizing and language can even be seen in the brain. Wernicke’s area, presumably the place where words are learned and stored, is right in the middle of the primate mentalizing regions. Indeed, the specific subarea of the left primate sensory cortex (called temporoparietal junction), which is highly selective for modeling other people’s intentions, knowledge, and beliefs, is entirely overlapping with Wernicke’s area—which is, as we have learned, required for people to understand speech and produce meaningful speech.
与此一致，儿童的心理化技能和语言技能是紧密相连的。在学龄前儿童中，语言技能的发展与心理化任务的表现之间存在显著的相关性，例如比如错误信念测试。影响心智化的疾病会导致类似的语言障碍。
Consistent with this, mentalizing skills and language skills in children are deeply interconnected. In preschoolers, there is a significant correlation between the development of language skills and performance in mentalizing tasks such as false-belief tests. Disorders that impair mentalizing cause similar impairments in language.
我们之所以能够操纵他人的思想，是因为语言似乎建立在我们内心模拟的直接窗口之上。听到句子会直接自动触发特定的心理意象。这就是为什么如果有人说了让我们心烦的话，我们不能简单地“不听”；我们必须捂住耳朵，否则无论我们是否喜欢，这些话都会直接触发模拟。
We are capable of puppeteering other minds because language is, it seems, built right on top of a direct window to our inner simulation. Hearing sentences directly and automatically triggers specific mental imagery. This is why if someone is saying something that upsets us, we cannot simply “not listen”; we must cover our ears, otherwise the words directly trigger simulations whether we like it or not.
但等一下...GPT- 4怎么样？
But Wait . . . What About GPT-4?
2023 年 3 月，OpenAI 发布了其全新升级的大型语言模型 GPT-4。GPT-4 的工作原理与 GPT-3 大致相同——它仅根据之前的单词序列来预测下一个单词。但与 GPT-3 不同的是，它是一个更大的模型，使用更多的数据进行训练。
In March 2023, OpenAI released GPT-4, its new and upgraded large language model. GPT-4 works in largely the same way as GPT-3—it predicts the next word based solely on the prior sequence of words. But unlike GPT-3, it is an even bigger model trained on even more data.
令人惊讶的是，我在本章中设计的每个问题都表明 GPT-3 缺乏常识和物理直觉，而 GPT-4 却完美地回答了这些问题。GPT-4 知道，如果你在地下室向上看，你会看到天花板而不是天空。GPT-4 知道，如果有人把棒球扔到你头顶一百英尺的地方，你是接不住的。GPT-4 甚至可以回答像 Sally-Ann 测试这样的心智理论问题。
Amazingly, each question that I designed in this chapter to demonstrate a lack of common sense and physical intuition in GPT-3 was answered flawlessly by GPT-4. GPT-4 knew that if you looked upward in a basement, you would see the ceiling not the sky. GPT-4 knew that if someone threw a baseball one hundred feet above your head, you would not be able to catch it. GPT-4 could even answer theory-of-mind questions like the Sally-Ann Test.
GPT-4 是如何做到的？OpenAI 没有添加内心世界模型或模拟他人思维的能力。相反，为了应对 GPT-3 在常识和推理问题上频繁犯错的情况，OpenAI开始专门针对常识和推理问题训练 GPT-4。他们使用了一种称为“从人类反馈中强化学习”的技术，当 GPT-4 回答错误时，人类会惩罚它，而当它回答正确时，人类会奖励它。他们甚至迫使 GPT-4 以特定的方式回答某些问题，以提高其性能。例如，OpenAI 通过写出每个步骤来训练 GPT-4 思考常识问题，这一技巧称为思路链提示。事实上，对比一下 GPT-3 和 GPT-4 分别如何回答球棒和球的问题：
How does GPT-4 do it? OpenAI did not add an inner world model or the ability to model the mind of others. Instead, in response to GPT-3’s frequent mistakes on commonsense and reasoning questions, OpenAI began training GPT-4 specifically on questions of commonsense and reasoning. They used a technique called “reinforcement learning from human feedback” in which humans punished GPT-4 when it answered questions incorrectly, and rewarded it when it answered correctly. They even pushed the GPT-4 to answer certain questions in specific ways to improve its performance. For example, OpenAI trained GPT-4 to think about commonsense questions by writing out each step, a trick called chain-of-thought prompting. Indeed, contrast how GPT-3 and GPT-4 each answer the bat and ball question:
我：一根球棒和一个球总共花了 1.10 美元。球棒比球贵 1.00 美元。球多少钱？
Me: A bat and a ball cost $1.10 in total. The bat costs $1.00 more than the ball. How much does the ball cost?
GPT-3： GPT-3: |
GPT-4： GPT-4: |
球的价格为 0.10 美元，因为如果球棒的价格比球贵 1.00 美元，而球棒和球的总价格为 1.10 美元，那么球的价格必定为 1.00 美元 - 1.10 美元 = 0.10 美元 The ball costs $0.10 because if the bat costs $1.00 more than the ball, and the bat and the ball cost $1.10 in total, then the ball must cost $1.00 - $1.10 = $0.10 （错误） (incorrect) |
我们将球的价格记为 x。那么球棒的价格就是 x + 1.00 美元，因为球棒的价格比球贵 1.00 美元。两件物品的总价格为 1.10 美元。因此，我们可以建立一个等式： Let’s denote the cost of the ball as x. Then, the cost of the bat would be x + $1.00, since it costs $1.00 more than the ball. The total cost of both items is $1.10. So, we can set up an equation: x (球) + x + 1.00 美元 (球棒) = 1.10 美元 x (ball) + x + $1.00 (bat) = $1.10 结合 x 的项，我们得到 Combining the terms for x, we get 2x + 1.00 美元 = 1.10 美元 2x + $1.00 = $1.10 现在，我们从两边减去 1.00 美元： Now, we’ll subtract $1.00 from both sides: 2x = 0.10 美元 2x = $0.10 最后，我们将两边都除以 2： Finally, we’ll divide both sides by 2: x = 0.05 美元 x = $0.05 所以，这个球的价格是0.05美元，也就是5美分。 So, the ball costs $0.05 or 5 cents. （正确的） (correct) |
通过训练 GPT-4 不仅能预测答案，还能预测推理答案的下一步，模型开始表现出思维的突现特性，但实际上并没有思考——至少不是像人类通过渲染世界的模拟来思考的方式。
By training GPT-4 to not just predict the answer, but to predict the next step in reasoning about the answer, the model begins to exhibit emergent properties of thinking, without, in fact, thinking—at least not in the way that a human thinks by rendering a simulation of the world.
尽管 GPT-4 正确回答了我在本章中概述的简单问题，但你仍然可以找到大量 GPT-4 失败的例子常识和心智理论问题。通过深入探究更复杂的问题，可以看出 GPT-4 缺乏世界模型。但找到这些例子正变得越来越困难。在某些方面，这已经成为一场打地鼠游戏；每当怀疑论者发布 LLM 回答错误的常识性问题的例子时，OpenAI 等公司就会将这些例子用作下次更新 LLM 的训练数据，从而正确回答这些问题。
Even though GPT-4 correctly answers the simpler questions I outlined in this chapter, you can still find plenty of examples of GPT-4 failing on commonsense and theory-of-mind questions. GPT-4’s lack of a world model can be seen by probing deeper with more complicated questions. But it is becoming progressively more painstaking to find these examples. In some ways, this has become a game of Whac-A-Mole; everytime a skeptic publishes examples of commonsense questions that LLMs answer incorrectly, companies like OpenAI simply use these examples as training data for the next update of their LLMs, which thereby answer such questions correctly.
事实上，这些模型的庞大规模以及训练它们所依据的天文数字般的数据量，在某种程度上掩盖了法学硕士和人类思维方式之间的根本差异。计算器的算术能力比任何人都强，但对数学的理解仍然不如人类。即使法学硕士正确回答了常识和心智理论问题，也不一定意味着它以相同的方式推理这些问题。
Indeed, the massive size of these models, along with the astronomical quantity of data on which they are trained, in some ways obscures the underlying differences between how LLMs think and how humans think. A calculator performs arithmetic better than any human, but still lacks the same understanding of math as a human. Even if LLMs correctly answer commonsense and theory-of-mind questions, it does not necessarily mean it reasons about these questions in the same way.
正如 Yann Lecun 所说，“LLM 的推理能力较弱，但其强大的联想记忆能力可以部分弥补这一缺陷。他们有点像死记硬背材料，但并没有真正建立起深厚的心理”事实上，这些法学硕士就像一台超级计算机，拥有巨大的内存容量，阅读的书籍和文章比一个人的大脑在一千次生命中所能消化的还要多。因此，看似常识性的推理实际上更像是模式匹配，是在天文数字般的庞大文本语料库上进行的。
As Yann Lecun said, “the weak reasoning abilities of LLMs are partially compensated by their large associative memory capacity. They are a bit like students who have learned the material by rote but haven’t really built deep mental models of the underlying reality.” Indeed, these LLMs, like a supercomputer, have a gargantuan memory capacity, having read more books and articles than a single human brain could consume in a thousand lifetimes. And so what seems like commonsense reasoning is really more like pattern matching, done over an astronomically enormous corpus of text.
但这些法学硕士仍然是一个了不起的进步。法学硕士的成功最令人惊奇之处在于，尽管他们只接受过语言方面的训练，但他们似乎对这个世界了解甚多。法学硕士可以正确地推理物理世界，而无需亲身体验这个世界。就像军事密码分析员解码加密秘密信息背后的含义，在原本胡言乱语中找到模式和含义一样，这些法学硕士只需扫描我们人类独特的思想传递代码的整个语料库，就能梳理出他们从未见过或听过、从未接触过或体验过的世界的各个方面。
But still, these LLMs are an incredible step forward. What is most amazing about the success of LLMs is how much they seemingly understand about the world despite being trained on nothing but language. LLMs can correctly reason about the physical world without ever having experienced that world. Like a military cryptanalyst decoding the meaning behind encryped secret messages, finding patterns and meanings in what was originally gibberish, these LLMs have been able to tease out aspects of a world they have never seen or heard, that they have never touched or experienced, by merely scanning the entire corpus of our uniquely human code for transferring thoughts.
有可能，或许是不可避免的，继续扩大这些通过为语言模型提供更多数据，这些模型将使它们在回答常识和心智理论问题时表现得更好。*但是，如果不结合外部世界的内部模型或其他思维模型——没有模拟和心智化的突破——这些 LLM 将无法捕捉人类智能的某些本质特征。而且 LLM 的采用速度越快——我们越需要它们做出决定——这些细微的差异就会变得越重要。
It is possible, perhaps inevitable, that continuing to scale up these language models by providing them with more data will make them even better at answering commonsense and theory-of-mind questions.* But without incorporating an inner model of the external world or a model of other minds—without the breakthroughs of simulating and mentalizing—these LLMs will fail to capture something essential about human intelligence. And the more rapid the adoption of LLMs—the more decisions we offload to them—the more important these subtle differences will become.
在人类大脑中，语言是我们内心模拟的窗口。语言是我们精神世界的界面。语言建立在我们模拟和推理他人思想的能力的基础上——推断他们的意思，并找出哪些词会在他们心中产生所需的模拟。我想大多数人都会同意，我们有一天会创造的类人人工智能不会是法学硕士；语言模型将仅仅是通往更丰富内涵的一扇窗户。
In the human brain, language is the window to our inner simulation. Language is the interface to our mental world. And language is built on the foundation of our ability to model and reason about the minds of others—to infer what they mean and figure out exactly which words will produce the desired simulation in their mind. I think most would agree that the humanlike artificial intelligences we will one day create will not be LLMs; language models will be merely a window to something richer that lies beneath.
突破 #5 总结：演讲
Summary of Breakthrough #5: Speaking
早期人类陷入了一场意想不到的完美风暴。非洲大草原的森林消亡将早期人类推向了制造工具、以肉食为生的境地，而这需要将工具的使用方法准确地传承给后代。原始语言的出现使工具的使用和制造技能得以成功地传承给后代。使语言得以产生的神经变化不是新的神经结构，而是对更古老结构的调整，这创造了一种语言学习程序；原始对话和共同注意的程序使儿童能够将名字与内在模拟的组成部分联系起来。通过这种课程训练，大脑皮层较旧的区域被重新用于语言。
Early humans got caught in an unlikely perfect storm of effects. The dying forests of the African savannah pushed early humans into a tool-making meat-eating niche, one that required the accurate propagation of tool use across generations. Proto-languages emerged, enabling tool use and manufacture skills to successfully propogate across generations. The neurological change that enabled language was not a new neurological structure but an adjustment to more ancient structures, which created a learning program for language; the program of proto-conversations and joint attention that enables children to tether names to components of their inner simulation. Trained with this curriculum, older areas of the neocortex were repurposed for language.
从此，人类开始尝试将这种原始语言用于不相关的个体，这引发了流言蜚语、利他主义和惩罚的反馈循环，不断选择更复杂的语言技能。随着社会群体的扩大和思想开始从一个大脑传到另一个大脑，人类的蜂巢思维应运而生，为思想的传播和代代积累创造了一个短暂的媒介。这需要更大的大脑来存储和分享更多积累的知识。也许正是由于这一点，或者是为了实现这一点，烹饪才被发明出来，它提供了大量的热量盈余，可以用来将大脑的体积扩大三倍。
From here, humans began experimenting with using this proto-language with unrelated individuals, and this kicked off a feedback loop of gossip, altruism, and punishment, which continuously selected for more sophisticated language skills. As social groups expanded and ideas began hopping from brain to brain, the human hive mind emerged, creating an ephemeral medium for ideas to propagate and accumulate across generations. This would have begged for bigger brains to store and share more accumulated knowledge. And perhaps due to this, or enabling it, cooking was invented, offering a huge caloric surplus that could be spent on tripling the size of brains.
因此，这场完美风暴催生了人类大脑进化史中的第五个也是最后一个突破：语言。语言带来了人类的许多独特特征，从利他主义到残忍。如果说有什么真正让人类与众不同的话，那就是思维不再是单一的，而是通过长期积累的思想与他人紧密相连。
And so, from this perfect storm emerged the fifth and final breakthrough in the evolutionary story of the human brain: language. And along with language came the many unique traits of humans, from altruism to cruelty. If there is anything that truly makes humans unique, it is that the mind is no longer singular but is tethered to others through a long history of accumulated ideas.
结论
Conclusion
第六次突破
The Sixth Breakthrough
随着现代人类大脑在大约十万年前在我们祖先身上的出现，我们四十亿年的进化史已经结束。回顾过去，我们可以开始勾勒出一幅画面——一个框架——人类大脑和智力出现的过程。我们可以将这个故事整合到我们的五大突破模型中。
WITH THE EMERGENCE of the modern human brain in our ancestors around one hundred thousand years ago, we have reached the conclusion of our four-billion-year evolutionary story. Looking back, we can begin to make out a picture—a framework—for the process by which the human brain and intelligence emerged. We can consolidate this story into our model of five breakthroughs.
第一个突破是转向：通过将刺激分为好刺激和坏刺激，转向好刺激和远离坏刺激，实现转向的突破。六亿年前，具有径向对称神经元的珊瑚状动物进化为具有双侧身体的动物。这些双侧身体计划将导航决策简化为二元转向选择；神经网络整合到第一个大脑中，使相反的价态信号能够整合成一个转向决策。多巴胺和血清素等神经调节剂使持续状态能够更有效地重新定位和局部搜索特定区域。联想学习使这些古老的蠕虫能够调整各种刺激的相对价态。在这个最初的大脑中，出现了动物的早期情感模板：快乐、痛苦、饱足和压力。
Breakthrough #1 was steering: the breakthrough of navigating by categorizing stimuli into good and bad, and turning toward good things and away from bad things. Six hundred million years ago, radially symmetric neuron-enabled coral-like animals reformed into animals with a bilateral body. These bilateral body plans simplified navigational decisions into binary turning choices; nerve nets consolidated into the first brain to enable opposing valence signals to be integrated into a single steering decision. Neuromodulators like dopamine and serotonin enabled persistent states to more efficiently relocate and locally search specific areas. Associative learning enabled these ancient worms to tweak the relative valence of various stimuli. In this very first brain came the early affective template of animals: pleasure, pain, satiation, and stress.
第二个突破是强化学习：学习重复历史上导致正效价的行为并抑制导致负效价的行为的突破。从人工智能的角度来看，这是无模型强化学习的突破。五亿年前，一个古老的两侧对称动物谱系长出了脊椎、眼睛、鳃和心脏，成为第一批脊椎动物，与现代鱼类最相似的动物。它们的大脑形成了所有现代脊椎动物的基本模板：皮层识别模式并构建空间地图，基底神经节通过反复试验进行学习。两者都建立在下丘脑中更古老的价机制遗迹之上。这种无模型强化学习带有一套熟悉的智力和情感特征：遗漏学习、时间感知、好奇心、恐惧、兴奋、失望和解脱。
Breakthrough #2 was reinforcing: the breakthrough of learning to repeat behaviors that historically have led to positive valence and inhibit behaviors that have led to negative valence. In AI terms, this was the breakthrough of model-free reinforcement learning. Five hundred million years ago, one lineage of ancient bilaterians grew a backbone, eyes, gills, and a heart, becoming the first vertebrates, animals most similar to modern fish. And their brains formed into the basic template of all modern vertebrates: the cortex to recognize patterns and build spatial maps and the basal ganglia to learn by trial and error. And both were built on top of the more ancient vestiges of valence machinery housed in the hypothalamus. This model-free reinforcement learning came with a suite of familiar intellectual and affective features: omission learning, time perception, curiosity, fear, excitement, disappointment, and relief.
第三个突破是模拟：在心理上模拟刺激和动作的突破。大约一亿年前，在一种四英寸长的祖先哺乳动物中，我们祖先脊椎动物的大脑皮层的部分区域转变为现代的大脑皮层。这种大脑皮层使动物能够在内部呈现对现实的模拟。这使它们能够在实际行动之前间接地向基底神经节展示该做什么。这是通过想象进行的学习。这些动物发展出了计划的能力。这使这些小型哺乳动物能够重新呈现过去的事件（情景记忆）并考虑过去的其他选择（反事实学习）。运动皮层后来的进化使动物不仅能够规划它们的整体导航路线，还能规划特定的身体动作，使这些哺乳动物拥有独特有效的精细运动技能。
Breakthrough #3 was simulating: the breakthrough of mentally simulating stimuli and actions. Sometime around one hundred million years ago, in a four-inch-long ancestral mammal, subregions of the cortex of our ancestral vertebrate transformed into the modern neocortex. This neocortex enabled animals to internally render a simulation of reality. This enabled them to vicariously show the basal ganglia what to do before the animal actually did anything. This was learning by imagining. These animals developed the ability to plan. This enabled these small mammals to re-render past events (episodic memory) and consider alternative past choices (counterfactual learning). The later evolution of the motor cortex enabled animals to plan not only their overall navigational routes but also specific body movements, giving these mammals uniquely effective fine motor skills.
第四项突破是心智化：即对自己心智进行建模的突破。大约一千万到三千万年前，早期灵长类动物的大脑皮层出现了新的区域，这些区域构建了早期哺乳动物大脑皮层区域的模型。实际上，这意味着这些灵长类动物不仅可以模拟动作和刺激（如早期哺乳动物），还可以模拟具有不同意图和知识的自己的心理状态。然后，这些灵长类动物可以应用该模型来预测自己未来的需求、理解他人的意图和知识（心智理论）以及通过观察学习技能。
Breakthrough #4 was mentalizing: the breakthrough of modeling one’s own mind. Sometime around ten to thirty million years ago, new regions of neocortex evolved in early primates that built a model of the older mammalian areas of neocortex. This, in effect, meant that these primates could simulate not only actions and stimuli (like early mammals), but also their own mental states with differing intent and knowledge. These primates could then apply this model to anticipating their own future needs, understanding the intents and knowledge of others (theory of mind), and learning skills through observation.
突破＃5 是说话：命名和语法的突破，将我们的内部模拟联系在一起，以实现跨代思想的积累。
Breakthrough #5 was speaking: the breakthrough of naming and grammar, of tethering our inner simulations together to enable the accumulation of thoughts across generations.
每一次突破都是因为先前的基础而成为可能。驾驶之所以成为可能，是因为神经元的进化强化学习之所以可能，只是因为它依靠已经进化的价神经元：没有价神经元，就没有开始强化学习的基础学习信号。模拟之所以可能，只是因为基底神经节中的试错学习已经存在。如果没有基底神经节来实现试错学习，就没有想象中的模拟影响行为的机制；通过让脊椎动物进化出真正的试错学习，哺乳动物中以后就可以出现替代试错。心理化之所以可能，只是因为模拟先于人类出现；心理化只是模拟哺乳动物大脑皮层的较老部分，同样的计算转向了内部。说话之所以可能，只是因为心理化先于人类出现；如果没有推断他人意图和知识的能力，你就无法推断要传达什么来帮助传达想法，也无法推断人们说话的意思。如果你没有推断他人知识和意图的能力，你就无法进行共同注意力这一关键步骤，即教师为学生识别物体。
Each breakthrough was possible only because of the building blocks that came prior. Steering was possible only because of the evolution of neurons earlier. Reinforcement learning was possible only because it bootstrapped on the valence neurons that had already evolved: without valence, there is no foundational learning signal for reinforcement learning to begin. Simulating was possible only because trial-and-error learning in the basal ganglia existed prior. Without the basal ganglia to enable trial-and-error learning, there would be no mechanism by which imagined simulations could affect behavior; by having actual trial-and-error learning evolve in vertebrates, vicarious trial and error could emerge later in mammals. Mentalizing was possible only because simulating came before; mentalizing is just simulating the older mammalian parts of the neocortex, the same computation turned inward. And speaking was possible only because mentalizing came before; without the ability the infer the intent and knowledge in the mind of another, you could not infer what to communicate to help transmit an idea or infer what people mean by what they say. And without the ability to infer the knowledge and intent of another, you could not engage in the crucial step of shared attention whereby teachers identify objects for students.
到目前为止，人类的故事是一部由两幕组成的传奇。第一幕是进化的故事：生物学上的现代人类如何从宇宙中原始的无生命物质中诞生。第二幕是文化的故事：社会上的现代人类如何从大约十万年前生物学上基本相同但文化上原始的祖先中诞生。
Thus far, humanity’s story has been a saga of two acts. Act 1 is the evolutionary story: how biologically modern humans emerged from the raw lifeless stuff of our universe. Act 2 is the cultural story: how societally modern humans emerged from largely biologically identical but culturally primitive ancestors from around one hundred thousand years ago.
虽然第一幕跨越了数十亿年，但我们在历史课上学到的大部分内容都是在第二幕这个相对较短的时间内展开的——所有文明、技术、战争、发现、戏剧、神话、英雄和恶棍都在这个时间窗口中展开，与第一幕相比，这只是一眨眼的功夫。
While act 1 unfolded over billions of years, most of what we have learned in history class unfolded during the comparatively much shorter time of act 2—all civilizations, technologies, wars, discoveries, dramas, mythologies, heroes, and villains unfolded in this time window that, compared to act 1, was a mere blink of an eye.
十万年前，一个智人的头脑里装着宇宙中最令人敬畏的物体之一；这是十亿多年艰苦的（即使无意的）进化过程的结果。她会舒适地坐在食物链的顶端，手里拿着长矛，穿着人造衣服取暖，驯服了火和无数的庞然大物，毫不费力地发挥出这么多智力壮举，完全不知道这些至今仍未被理解的能力是如何产生的，当然，也不知道同时这段伟大、悲惨而又美妙的旅程最终将在她的智人后代身上展开。
An individual Homo sapiens one hundred thousand years ago housed in her head one of the most awe-inspiring objects in the universe; the result of over a billion years of hard—even if unintentional—evolutionary work. She would have sat comfortably at the top of the food chain, spear in hand, warmed in manufactured clothing, having tamed both fire and countless gargantuan beasts, effortlessly invoking these many intellectual feats, utterly unaware of the past by which these still yet-to-be-understood abilities came to be and also, of course, unaware of the simultaneously magnanimous, tragic, and wonderful journey that would eventually unfold in her Homo sapiens descendants.
所以你现在正在读这本书。几乎不可能的大量事件导致了这一刻的到来：热液喷口中第一批冒泡的细胞；单细胞生物的第一次捕食战斗；多细胞生物的诞生；真菌和动物的分化；祖先珊瑚中第一批神经元和反射的出现；古代两侧对称动物中第一批具有效价和情感以及联想学习的大脑的出现；脊椎动物的兴起和对时间、空间、模式和预测的驯服；躲避恐龙的微小哺乳动物模拟的诞生；树栖灵长类动物的政治建构和心理化；早期人类语言的出现；当然，还有过去数十万年来在数十亿具有语言能力的人类大脑中积累的无数想法的创造、修改和毁灭。这些想法已经积累到了现代人类可以在电脑上打字、写字、使用手机、治病的地步，甚至可以根据我们的形象构建新的人工智能。
And so here you are, reading this book. An almost impossibly vast number of events led to this exact moment: the first bubbling cells in hydrothermal vents; the first predatory battles of single-celled organisms; the birth of multicellularity; the divergence of fungi and animals; the emergence of the first neurons and reflexes in ancestral corals; the emergence of the first brains with valence and affect and associative learning in ancient bilaterians; the rise of vertebrates and the taming of time, space, patterns, and prediction; the birth of simulation in minuscule mammals hiding from dinosaurs; the construction of politics and mentalizing of tree-living primates; the emergence of language in early humans; and, of course, the creation, modification, and destruction of countless ideas that have accumulated throughout the billions of language-enabled human brains over the past hundreds of thousands of years. These ideas have accumulated to the point that modern humans can now type on computers, write words, use cell phones, cure diseases, and, yes, even construct new artificial intelligences in our image.
进化仍在认真展开；我们还没有走到智慧故事的尽头，而是刚刚开始。地球上的生命只有 40 亿年的历史。太阳还要再过 70 亿年才会灭亡。因此，至少地球上的生命还有 70 亿年左右的时间来培育新的生物智慧形式。如果地球上的原始分子只需要 45 亿年就能转化为人类大脑，那么在 70 亿年的进化中，智慧能走多远？假设生命以某种方式走出了太阳系，或者至少生命独立地出现在宇宙的其他地方，那么进化将有天文数字般的时间开始：宇宙膨胀到新恒星停止形成还需要一万亿年以上，最后一个星系分裂还需要一千万亿年。我们很难想象我们这个拥有 140 亿年历史的宇宙到底有多年轻。如果将宇宙的千万亿年时间轴压缩为一个日历年，那么我们会发现，今天只处于一年中的前七分钟，甚至还不是第一天的黎明。
Evolution is still unfolding in earnest; we are not at the end of the story of intelligence but at the very beginning. Life on Earth is only four billion years old. It will be another seven billion years before our sun dies. And thus life, at least on Earth, has another seven or so billion years to tinker with new biological forms of intelligence. If it took only four and a half billion years for raw molecules on Earth to transform into human brains, how far can intelligence get in another seven billion years of evolution? And assuming life does, somehow, make its way out of the solar system, or at least life independently shows up elsewhere in the universe, there will be astronomically more time for evolution to get to work: it will be over a trillion years before the universe has expanded so greatly that new stars cease to form, and a quadrillion before the last galaxy breaks apart. It can be hard to conceptualize just how young our fourteen-billion-year-old universe actually is. If you took the quadrillion-year timeline of our universe and squished it into a single calendar year, then we would find ourselves, today, at only the first seven minutes of the year, not even at the dawn of the very first day.
如果我们对物理学的现代理解是正确的，那么从现在起大约一千万亿年后，当最后一个星系最终分裂开来时，宇宙将开始缓慢地衰落，毫无意义地进入不可避免的热寂。这是熵不可阻挡的趋势所导致的不幸结果，熵是宇宙中不可阻挡的原始力量，40 亿年前，第一批自我复制的 DNA 分子开始与之抗争。通过自我复制，DNA 摆脱了熵，不再以物质而是以信息的形式存在。第一串 DNA 之后的所有进化创新都秉承着这种精神，坚持不懈，反击熵，拒绝消失。在这场伟大的战斗中，通过语言在人类大脑之间传播的思想是生命的最新创新，但肯定不会是最后一次。我们仍然在山脚下，只是在通往某个目标的长阶梯的第五级台阶上。
If our modern understanding of physics is correct, then about a quadrillion years from now, after the the last galaxy has finally broken apart, the universe will begin its slow process of fading meaninglessly into an inevitable heat death. This is the unfortunate result of the inexorable trend of entropy, that raw unstoppable force of the universe that the first self-replicated DNA molecules began their war against four billion years ago. By self-replicating, DNA finds respite from entropy, persisting not in matter but in information. All the evolutionary innovations that followed the first string of DNA have been in this spirit, the spirit of persisting, of fighting back against entropy, of refusing to fade into nothingness. And in this great battle, ideas that float from human brain to human brain through language are life’s newest innovation but will surely not be its last. We are still at the base of the mountain, only on the fifth step on a long staircase to something.
当然，我们不知道第六个突破会是什么。但第六个突破似乎越来越有可能是超级人工智能的诞生；我们的后代在硅片中出现，按照我们的形象制造的智能从生物媒介转变为数字媒介。从这种新媒介中，单一智能的认知能力将出现天文数字般的扩张。人类大脑的认知能力受到神经元处理速度、人体热量限制以及大脑在碳基生命形式中可以达到多大尺寸的限制。当智能摆脱这些生物限制时，就会出现第六个突破。硅基人工智能可以根据需要无限扩大其处理能力。事实上，随着人工智能可以自由复制和重新配置自身，个性将失去其明确的界限；随着生物交配机制让位于新的硅基训练和创造新智能实体的机制，父母身份将具有新的意义。甚至进化本身也将被抛弃，至少是它熟悉的形式；智能将不再受制于缓慢的基因变异和自然选择过程，而是受制于更基本的进化原理，即最纯粹意义上的变异和选择——随着人工智能不断自我改造，那些选择支持更好生存的特征的人工智能当然会成为最终生存下来的人工智能。
Of course, we don’t know what breakthrough #6 will be. But it seems increasingly likely that the sixth breakthrough will be the creation of artificial superintelligence; the emergence of our progeny in silicon, the transition of intelligence—made in our image—from a biological medium to a digital medium. From this new medium will come an astronomical expansion in the scale of a single intelligence’s cognitive capacity. The cognitive capacity of the human brain is hugely limited by the processing speed of neurons, the caloric limitations of the human body, and the size constraints of how big a brain can be and still fit in a carbon-based life-form. Breakthrough #6 will be when intelligence unshackles itself from these biological limitations. A silicon-based AI can infinitely scale up its processing capacity as it sees fit. Indeed, individuality will lose its well-defined boundaries as AIs can freely copy and reconfigure themselves; parenthood will take on new meaning as biological mechanisms of mating give way to new silicon-based mechanisms of training and creating new intelligent entities. Even evolution itself will be abandoned, at least in its familiar form; intelligence will no longer be entrapped by the slow process of genetic variation and natural selection, but instead by more fundamental evolutionary principles, the purest sense of variation and selection—as AIs reinvent themselves, those who select features that support better survival will, of course, be the ones that survive.
无论下一步发展出哪种智力策略，它们肯定都会包含它们起源的人类智能的痕迹。虽然这些人工智能的底层媒介可能不保留大脑的任何生物包袱，但这些实体仍将不可逆转地建立在之前的五项突破的基础上。这不仅是因为这五项突破是人类创造者智能的基础——创造者不禁会将自己的痕迹注入到他们的创造物中——还因为它们至少在最初将被设计为与人类互动，从而将植入人类智能的重现，或至少是镜像。
And whichever intellectual strategies end up evolving next, they will surely contain hints of the human intelligence from which they came. While the underlying medium of these artificial superintelligences may retain none of the biological baggage of brains, these entities will still irrevocably be built on the foundation of the five breakthroughs that came before. Not only because these five breakthroughs were the foundation of the intelligence of their human creators—creators cannot help but imbue their creations with hints of themselves—but also because they will be designed, at least at first, to interact with humans, and thereby will be seeded with a recapitulation, or at least a mirror, of human intelligence.
因此，我们正站在人类智能史上第六次突破的悬崖边上，即将掌握生命诞生的过程，诞生超级智能的人工智能。在这个悬崖边上，我们面临着一个非常不科学但实际上更为重要的问题：人类的目标应该是什么？这不是真理问题，而是价值观问题。
And so we stand on the precipice of the sixth breakthrough in the story of human intelligence, at the dawn of seizing control of the process by which life came to be and of birthing superintelligent artificial beings. At this precipice, we are confronted with a very unscientific question but one that is, in fact, far more important: What should be humanity’s goals? This is not a matter of veritas—truth—but of values.
正如我们所见，过去的选择会随着时间而传播。因此，我们如何回答这个问题将对未来产生影响。我们会遍布整个星系吗？探索宇宙的隐藏特征，构建新的思想，解开宇宙的秘密，发现意识的新特征，变得更加富有同情心，参与难以想象的冒险？还是我们会失败？我们的骄傲、仇恨、恐惧和部落主义的进化包袱会将我们撕裂吗？我们是否会成为又一次以悲剧收场的进化迭代？也许在人类灭绝数百万年后，地球上的一些较晚出现的物种会再次尝试迈出下一步——也许是倭黑猩猩、章鱼、海豚或波西亚蜘蛛。也许他们会像我们发现恐龙化石一样发现我们的化石，思考我们曾经的生活，并撰写有关我们大脑的书籍。甚至更糟的是，也许我们人类将通过破坏地球的气候或用核战争将我们的世界摧毁，从而结束地球上四十亿年的伟大生命实验。
As we have seen, past choices propagate through time. And so how we answer this question will have consequences for eons to come. Will we spread out across galaxies? Explore the hidden features of the cosmos, construct new minds, unravel the secrets of the universe, find new features of consciousness, become more compassionate, engage in adventures of unthinkable scope? Or will we fail? Will our evolutionary baggage of pride, hatred, fear, and tribalism rip us apart? Will we go down as just another evolutionary iteration that came to a tragic end? Perhaps it will be some later species on Earth, millions of years after humans have gone extinct, that will make another stab at taking the next step up the mountain—perhaps the bonobos or octopuses or dolphins or Portia spiders. Perhaps they will uncover our fossils as we have uncovered those of dinosaurs and ponder what lives we must have lived and write books about our brains. Or even worse, perhaps we humans will end the grand four-billion-year experiment of life on Earth through ravaging the planet’s climate or blowing our world into oblivion with nuclear warfare.
当我们展望这个新时代时，我们有必要回顾一下大脑诞生的漫长十亿年历程。随着我们被赋予神一般的创造能力，我们应该从中学习上帝——不加思考的进化过程——在我们之前就已经存在。我们对自己的思想了解得越多，我们就越有能力按照自己的形象创造人造思想。我们对思想形成过程了解得越多，我们就越有能力选择要丢弃哪些智能特征，要保留哪些智能特征，以及要改进哪些智能特征。
As we look forward into this new era, it behooves us to look backward at the long billion-year story by which our brains came to be. As we become endowed with godlike abilities of creation, we should learn from the god—the unthinking process of evolution—that came before us. The more we understand about our own minds, the better equipped we are to create artificial minds in our image. The more we understand about the process by which our minds came to be, the better equipped we are to choose which features of intelligence we want to discard, which we want to preserve, and which we want to improve upon.
我们是这场历时 140 亿年的伟大变革的中坚力量。无论我们是否愿意，宇宙已经把接力棒交给了我们。
We are the stalwarts of this grand transition, one that has been fourteen billion years in the making. Whether we like it or not, the universe has passed us the baton.
致谢
Acknowledgments
写这本书是人类慷慨的一个案例。这本书之所以能够问世，是因为有许多人非常善良，他们帮助我完成了这本书。有很多人值得感谢。
Writing this book was a case study in human generosity. It was only possible because of the remarkable kindness of many people who helped me bring it life. There are many people who deserve thanks.
首先，我要感谢我的妻子悉尼，她编辑了许多页面，并帮助我思考了许多概念上的障碍。无数个早晨，她醒来时发现我早已不见踪影，因为我已经溜出去读书写作了；无数个日子，她下班回家时发现我躲在办公室里。感谢你支持这项工作，尽管它占用了太多的脑力空间。
First and foremost, my wife, Sydney, who edited many pages and helped me think through many conceptual snags. She woke up countless mornings to find me long gone because I had already snuck out to read and write; and she came home from work countless days to find me tucked away in my office. Thank you for supporting this endeavor despite how much mental space it consumed.
我要感谢我的第一批读者，他们给了我反馈和鼓励：乔纳森·巴尔科姆 (Jonathan Balcome)、杰克·贝内特 (Jack Bennett)、琪琪·弗里德曼 (Kiki Freedman)、马库斯·杰克林 (Marcus Jecklin)、达娜·纳贾尔 (Dana Najjar)、吉迪恩·科瓦德罗 (Gideon Kowadlo)、法耶兹·穆罕默德 (Fayez Mohamood)、沙玛拉·雷迪 (Shyamala Reddy)、比利·斯坦 (Billy Stein)、安珀·图内尔 (Amber Tunnell)、迈克尔·韦斯 (Michael Weiss)、马克斯·温尼克 (Max Wenneker)，当然还有我的父母加里·贝内特 (Gary Bennett) 和凯西·克罗斯特 (Kathy Crost)；以及我的继母艾丽莎·贝内特 (Alyssa Bennett)。
I want to thank my initial readers, who gave me feedback and encouragement: Jonathan Balcome, Jack Bennett, Kiki Freedman, Marcus Jecklin, Dana Najjar, Gideon Kowadlo, Fayez Mohamood, Shyamala Reddy, Billy Stein, Amber Tunnell, Michael Weiss, Max Wenneker, and, of course, my parents, Gary Bennett and Kathy Crost; and my stepmother, Alyssa Bennett.
我尤其要感谢我的岳父 Billy Stein，他对人工智能或神经科学没有内在兴趣，但还是尽职尽责地阅读和注释了每一页，质疑每一个概念和想法以确保其合理，并在结构、可理解性和流程方面提供了宝贵的意见和指导。Dana Najjar、Shyamala Reddy 和 Amber Tunnell 的写作经验比我丰富得多，他们为我提供了早期草稿的重要意见。还有 Gideon Kowaldo，他为我提供了有关人工智能历史和概念的有用意见。
In particular, I want to thank my father-in-law, Billy Stein, who has no intrinsic interest in AI or neuroscience, but nonetheless dutifully read and annotated every single page, questioned every concept and idea to make sure it made sense, and provided invaluable input and guidance on structure, understandability, and flow. Dana Najjar, Shyamala Reddy, and Amber Tunnell, who have far more writing experience than I, gave me essential input on early drafts. And Gideon Kowaldo, who gave me useful input on the AI history and concepts.
我非常感谢在百忙之中抽出时间的科学家们他们热心回复我的电子邮件，我向他们提出了无数问题。他们帮助我理解他们的研究，并思考本书中的许多概念：Charles Abramson、Subutai Ahmed、Bernard Balleine、Kent Berridge、Culum Brown、Eric Brunet、Randy Bruno、Ken Cheng、Matthew Crosby、Francisco Clasca、Caroline DeLong、Karl Friston、Dileep George、Simona Ginsburg、Sten Grillner、Stephen Grossberg、Jeff Hawkins、Frank Hirth、Eva Jablonka、Kurt Kotrschal、Matthew Larkum、Malcolm MacIver、Ken-ichiro Nakajima、Thomas Parr、David Redish、Murray Sherman、James Smith 和 Thomas Suddendorf。如果没有他们愿意回答陌生人的问题，像我这样的人不可能学习一个新领域。
I am extremely grateful to the scientists who took time out of their busy lives to respond to my emails where I peppered them with innumerable questions. They helped me understand their research and think through many of the concepts in this book: Charles Abramson, Subutai Ahmed, Bernard Balleine, Kent Berridge, Culum Brown, Eric Brunet, Randy Bruno, Ken Cheng, Matthew Crosby, Francisco Clasca, Caroline DeLong, Karl Friston, Dileep George, Simona Ginsburg, Sten Grillner, Stephen Grossberg, Jeff Hawkins, Frank Hirth, Eva Jablonka, Kurt Kotrschal, Matthew Larkum, Malcolm MacIver, Ken-ichiro Nakajima, Thomas Parr, David Redish, Murray Sherman, James Smith, and Thomas Suddendorf. Without their willingness to respond to the questions of a complete stranger, it would have been impossible for someone like me to learn a new field.
我要特别感谢 Karl Friston、Jeff Hawkins 和 Subutai Ahmed，他们阅读了我早期的一些论文，并慷慨地对我进行指导，并带我到他们的实验室分享我的想法并向他们学习。
I want to especially thank Karl Friston, Jeff Hawkins, and Subutai Ahmed, who read some of my early papers and generously took me under their wing and brought me into their labs to share my ideas and learn from them.
Joseph LeDoux、David Redish 和 Eva Jablonka 非常慷慨地抽出时间。他们不仅阅读并注释了手稿的多个草稿，还对我遗漏的概念、我未考虑的文献领域提供了重要反馈，并帮助我扩展了框架和故事。他们成了我事实上的神经科学编辑和顾问。本书中被认为有价值的任何方面都应归功于他们（而被认为不有价值的方面不应受到指责）。
Joseph LeDoux, David Redish, and Eva Jablonka were astoundingly generous with their time. Not only did they read and annotate multiple drafts of the manuscript, but they provided essential feedback on concepts I had missed, areas of the literature I had failed to consider, and helped me expand on the framework and story. They became my de facto neuroscience editors and advisers. They deserve much of the credit for whatever aspects of this book are deemed valuable (and none of the blame for aspects deemed otherwise).
这本书我最喜欢的部分之一是艺术，而 Rebecca Gelernter 和 Mesa Schumacher 功不可没。他们是才华横溢的艺术家，创作了本书中精美的艺术作品。
One of my favorite parts of this book is the art, and for this, Rebecca Gelernter and Mesa Schumacher deserve all the credit. They are the incredibly talented artists who produced the beautiful art herein.
作为一名首次写作的作家，我很感谢图书行业为我提供指导的人们。简·弗里德曼 (Jane Friedman) 给了我严厉而有用的反馈。作家乔纳森·巴尔科姆 (Jonathan Balcome) 阅读了最早的草稿之一，并给予了反馈和鼓励。作家格里·赫尔希 (Gerri Hirshey) 和杰米·卡尔 (Jamie Carr) 分别帮助我制定了图书提案，并对早期章节给予了反馈。
As a first-time author, I am grateful to the people in the book industry who gave me guidance. Jane Friedman gave me tough and useful feedback. The writer Jonathan Balcome read one of the earliest drafts and gave feedback and encouragement. The writers Gerri Hirshey and Jamie Carr each helped me with my book proposal and gave me feedback on early chapters.
哈珀柯林斯的丽莎·夏基 (Lisa Sharkey) 让这本书成为现实。在我决定写这本书之前，我和她谈过，问她，鉴于我是第一次写书，而不是受过正式训练的神经科学家，是否值得尝试写这本书。尽管很明显，这本书很有可能这本书无法问世，她鼓励我无论如何都要坚持下去。我非常感谢那次谈话，以及她的建议和支持。在那次谈话一年多之后，她最终决定出版这本书，这真是太合适了。
Lisa Sharkey at HarperCollins made this book real. I spoke to her before I decided to write it and asked her whether it was even worth attempting to write this book given I was a first-time author and not a formally trained neuroscientist. Despite the obvious fact that there was a good chance the book wouldn’t see the light of day, she encouraged me to pursue it regardless. I am deeply grateful for that conversation, and her advice and support. It is wonderfully fitting that she was the one, over a year after that conversation, who ended up deciding to publish this book.
我要感谢我的经纪人吉姆·莱文 (Jim Levine)，他愿意只看一个介绍就读完这本书（感谢杰夫·霍金斯 (Jeff Hawkins)）。吉姆在一天内读完了整本书，第二天就下注了。我要感谢我的美国编辑马特·哈珀 (Matt Harper) 和我的英国编辑迈尔斯·阿奇博尔德 (Myles Archibald)，他们也对这本书下了注，帮助我完成了无数的草稿，并指导我度过写作过程中的起起伏伏。我要感谢我的文字编辑特雷西·罗 (Tracy Roe)，她有条不紊地纠正了我的许多拼写错误和语法错误。
I want to thank my agent, Jim Levine, who was willing to read the book from nothing but a single introduction (thanks to Jeff Hawkins). Jim read the entire book in one day, and took a bet on it the next day. I want to thank my U.S. editor, Matt Harper, and my U.K. editor, Myles Archibald, who also took a bet on this book, and helped me work through countless drafts and navigate the many ups and downs of writing. I want to thank my copyeditor, Tracy Roe, who methodically fixed my many typos and grammatical mishaps.
还有一些人以不那么直接但同样重要的方式帮助了我。我的吉他老师 Stephane Wrembel，我曾多次向他寻求建议。我的朋友 Ally Sprague（她往往兼任我的教练）帮助我决定休假一年来写这本书。我的朋友 Dougie Gliecher 和 Ben Eisenberg，他们把我介绍给他们在图书行业认识的人。我的兄弟 Adam Bennett 和 Jack Bennett，他们给我的生活带来欢乐和乐趣，并且总是我的灵感源泉。还有我的父母 Gary Bennett 和 Kathy Crost，他们培养了我对学习的热爱，向我展示了如何追随我的好奇心，并教我完成我开始的事情。
There are also folks who helped me in less direct but equally important ways. My guitar teacher, Stephane Wrembel, who I turned to for advice on numerous occasions. My friend Ally Sprague (who tends to double as my coach), who helped me make the decision to take a year off to write this book. My friends Dougie Gliecher and Ben Eisenberg, who connected me to people they knew in the book industry. My brothers, Adam Bennett and Jack Bennett, who bring joy and play to my life, and are always a source of inspiration. And my parents, Gary Bennett and Kathy Crost, who fostered in me a love of learning, showed me how to follow my curiosity, and taught me to finish things I start.
本书之所以能够问世，是因为之前的许多作品，它们的思想、故事和写作方式从根本上塑造了本书。《对齐问题》作者：布莱恩·克里斯蒂安。《行为举止》作者：罗伯特·萨波斯基。《我们自己的深层历史》作者：约瑟夫·勒杜。《敏感灵魂的进化》作者：伊娃·雅布隆卡和西蒙娜·金斯伯格。《猴子如何看世界》作者：多萝西·切尼和罗伯特·西法斯。《大脑中的思维》作者：大卫·雷迪什。《论智力》和《千个品牌》作者：杰夫·霍金斯。《为什么只有我们》作者：罗伯特·贝维什和诺姆·乔姆斯基。
This book was only possible because of many other prior works whose ideas, stories, and writing shaped this book in fundamental ways. The Alignment Problem by Brian Christian. Behave by Robert Sapolsky. The Deep History of Ourselves by Joseph LeDoux. The Evolution of the Sensitive Soul by Eva Jablonka and Simona Ginsburg. How Monkeys See the World by Dorothy Cheney and Robert Seyfarth. The Mind within the Brain by David Redish. On Intelligence and A Thousand Brands by Jeff Hawkins. Why Only Us by Robert Berwish and Noam Chomsky.
还有许多教科书成为我必不可少的资源。Georg Striedter 和 R. Glenn Northcutt 的《穿越时空的大脑》。Gerald Schneider 的《大脑结构及其起源》 。Ian Goodfellow、Yoshua Bengio 和 Aaron Courville 的《深度学习》 。Jon H. Kaas 的《进化神经科学》。W. Tecumseh Fitch 的《语言的进化》 。鱼《认知与行为》，作者：Culum Brown、Kevin Laland 和 Jens Krause。《神经经济学》，作者：Paul Glimcher。《前额皮质的神经生物学》，作者：Richard Passingham 和 Steven Wise。《新高管大脑》，作者：Elkhonon Goldberg。《强化学习》，作者：Richard Sutton 和 Andrew Barto。
There were also numerous textbooks that became essential resources for me. Brains Through Time by Georg Striedter and R. Glenn Northcutt. Brain Structure and Its Origins by Gerald Schneider. Deep Learning by Ian Goodfellow, Yoshua Bengio, and Aaron Courville. Evolutionary Neuroscience by Jon H. Kaas. The Evolution of Language by W. Tecumseh Fitch. Fish Cognition and Behavior by Culum Brown, Kevin Laland, and Jens Krause. Neuroeconomics by Paul Glimcher. The Neurobiology of the Prefrontal Cortex by Richard Passingham and Steven Wise. The New Executive Brain by Elkhonon Goldberg. Reinforcement Learning by Richard Sutton and Andrew Barto.
最后，我要感谢我的狗查理，它乞求零食和嬉戏地推我，迫使我从无数次睡眼惺忪的阅读报纸和教科书的时光中重新回到现实生活。在我写这段话的时候，它正躺在我旁边熟睡，在梦中抽搐着，它的大脑皮层肯定在模拟某种东西。当然，我永远不会知道是什么。
Lastly, I want to thank my dog, Charlie, whose begging for treats and playful nudges forced me to reenter the world of the living from numerous bleary-eyed sessions of reading papers and textbooks. As I write this paragraph, she is lying next to me fast asleep, twitching away in some dream, her neocortex surely rendering a simulation of something. Of what, of course, I will never know.
词汇表
Glossary
习得（与联想学习有关）：基于新的经验，在刺激和反应之间形成新的关联的过程（即“习得”）
acquisition (in relation to associative learning): the process by which a new association between a stimulus and a response is formed (i.e., “acquired”) based on new experience
联想学习：将刺激与反射反应联系起来的能力，这样下次出现该刺激时，更有可能发生相同的反射反应
associative learning: the ability to associate a stimulus with a reflexive response, such that the next time that stimulus occurs that same reflexive response is more likely to occur
适应（与神经元的反应有关）：神经元的属性，即它们会改变给定刺激强度与由此产生的放电率之间的关系；例如，随着时间的推移，神经元会逐渐降低其放电率以响应恒定的刺激
adaptation (in relation to the responses of neurons): the property of neurons whereby they change the relationship between a given stimulus strength and the resulting firing rate; for example, neurons will gradually decrease their firing rate in response to a constant stimulus over time
情感/情绪状态：一种根据效价（正效价或负效价）和唤醒（高唤醒或低唤醒）对动物行为状态进行分类的方法
affect/affective state: a way to categorize the behavioral state of an animal along the dimensions of valence (either positive valence or negative valence) and arousal (either high arousal or low arousal)
无颗粒前额皮质 (aPFC)：早期哺乳动物进化而来的额叶新皮质区域。它之所以被称为“无颗粒”，是因为它是缺少第 4 层（包含“颗粒细胞”的层）的新皮质区域
agranular prefrontal cortex (aPFC): the region of frontal neocortex that evolved in early mammals. It is called “agranular” because it is a region of neocortex that is missing layer 4 (the layer that contains “granule cells”)
自动联想：某些神经元网络的一种特性，神经元会自动与自身建立关联，使得网络在给定不完整模式时能够自动完成模式
auto-association: a property of certain networks of neurons whereby neurons automatically build associations with themselves, enabling the network to automatically complete patterns when given an incomplete pattern
反向传播：一种训练人工神经网络的算法；计算改变给定连接的权重对网络末端误差（实际输出与期望输出之间的差异的度量）的影响，并相应地调整每个权重以减少误差
backpropagation: an algorithm for training artificial neural networks; computes the impact of changing the weight of a given connection on the error (a measure of the difference between the actual output and the desired output) at the end of the network, and nudges each weight accordingly to reduce the error
两侧对称动物：约 6 亿年前拥有共同祖先的一群物种，其中出现了两侧对称性以及第一个大脑
bilaterian: a group of species with a common ancestor around 600 million years ago, in whom bilaterial symmetry emerged as well as the first brains
双侧对称：动物体只有一个对称面，将动物大致分为左右两半，两半互为镜像
bilateral symmetry: animal bodies that contain a single plane of symmetry, which divides the animal into roughly mirror image right and left halves
阻断（与联想学习有关）：早期两侧对称动物进化而来的信用分配问题的解决方案之一；一旦动物在预测线索和反应之间建立了关联，所有与预测线索重叠的进一步线索都会受到抑制（即“阻断”），无法与该反应建立关联
blocking (in relation to associative learning): one of the solutions to the credit assignment problem that evolved in early bilaterians; once an animal has established an association between a predictive cue and a response, all further cues that overlap with the predictive cue are inhibited (i.e., “blocked”) from making associations with that response
灾难性遗忘：按顺序训练神经网络（而不是一次性训练所有神经网络）的一个突出挑战；当你教神经网络识别新模式时，它往往会丢失以前学过的旧模式的记忆
catastrophic forgetting: an outstanding challenge of sequentially training neural networks (as opposed to training them all at once); when you teach a neural network to recognize new patterns, it tends to lose the memory of previously learned old patterns
持续学习：随着新数据的提供，自动学习和记住新事物的能力
continual learning: the ability to automatically learn and remember new things as new data is provided
卷积神经网络：一种神经网络，旨在通过在不同位置寻找相同特征来识别图像中的物体
convolutional neural network: a type of neural network designed to recognize objects in images by looking for the same features in different locations
信用分配问题：当发生某个事件或结果时，您会给予什么线索或行动“信用”来预测该事件或结果？
credit assignment problem: when an event or outcome occurs, what cue or action do you give “credit” for being predictive of that event or outcome?
消退（与联想学习有关）：由于条件刺激不再与后续反射反应同时发生（即，以前在食物前发生的蜂鸣器响起，但现在不再在食物前发生），导致先前学到的联想受到抑制（即“消退”）的过程
extinction (in relation to associative learning): the process by which previously learned associations are inhibited (i.e., “extinguished”) due to a conditional stimulus no longer occurring alongside a subsequent reflexive response (i.e., a buzzer sounding that used to occur before food, but no longer occurs before food)
发放率（也称脉冲率）：神经元每秒产生的脉冲数
firing rate (also spike rate): the number of spikes per second generated by a neuron
生成模型：一种概率模型，它学习生成自己的数据，并通过将生成的数据与实际数据进行比较来识别事物（一些研究人员称之为“推理感知”的过程）
generative model: a type of probabilistic model that learns to generate its own data, and recognizes things by comparing generated data with actual data (a process some researchers call “perception by inference”)
颗粒前额皮质 (gPFC)：早期灵长类动物进化而来的额叶新皮质区域。它之所以被称为“颗粒”，是因为它是前额叶新皮质的一个区域，其中包含第 4 层（包含“颗粒细胞”的层）
granular prefrontal cortex (gPFC): the region of frontal neocortex that evolved in early primates. It is called “granular” because it is a region of prefrontal neocortex that contains a layer 4 (the layer that contains “granule cells”)
亥姆霍兹机器：亥姆霍兹通过推理感知思想的早期证明
Helmholtz machine: an early proof of concept of Helmholtz’s idea of perception by inference
心理化：对自己内心模拟进行模拟的行为（即思考自己的思维）
mentalizing: the act of rendering a simulation of one’s own inner simulation (i.e., thinking about your own thinking)
基于模型的强化学习：一种强化学习，在选择动作之前，提前“演绎”（即模拟）可能的未来动作
Model-based reinforcement learning: the type of reinforcement learning whereby possible future actions are “played out” (i.e., simulated) ahead of time before selecting an action
无模型强化学习：一种强化学习，其中可能的未来行动不会提前“演绎”（即模拟）；而是基于当前情况自动选择行动
model-free reinforcement learning: the type of reinforcement learning whereby possible future actions are not “played out” (i.e., simulated) ahead of time; instead, actions are automatically selected based on the current situation
神经调节剂：一些神经元（“神经调节神经元”）释放的一种化学物质，对许多下游神经元具有复杂且通常持久的影响。著名的神经调节剂包括多巴胺、血清素和肾上腺素
neuromodulator: a chemical released by some neurons (“neuromodulatory neurons”) that has complex and often long-lasting effects on many downstream neurons. Famous neuromodulators include dopamine, serotonin, and adrenaline
掩盖（与联想学习有关）：早期两侧对称动物进化而来的信用分配问题的解决方案之一；当动物有多个预测线索可供使用时，它们的大脑倾向于选择最强的线索（即强线索掩盖了弱线索）。
overshadowing (in relation to associative learning): one of the solutions to the credit assignment problem that evolved in early bilaterians; when animals have multiple predictive cues to use, their brains tend to pick the cues that are the strongest (i.e., strong cues overshadow weak cues).
灵长类感觉皮层 (PSC)：早期灵长类动物进化而来的感觉新皮层的新区域，包括颞上沟 (STS) 和颞顶交界处 (TPJ)
primate sensory cortex (PSC): the new regions of sensory neocortex that evolved in early primates, these include the superior temporal sulcus (STS) and temporoparietal junction (TPJ)
重新习得（与联想学习有关）：应对世界上不断变化的突发事件，并使早期的两侧对称动物能够不断学习；旧的联系比全新的联系更容易重新获得
reacquisition (in relation to associative learning): one of the techniques to deal with changing contingencies in the world and enable continual learning in early bilaterians; old-extinguished associations are reacquired faster than entirely new associations
感觉新皮质：新皮质的后半部分，即模拟外部世界的区域
sensory neocortex: the back half of the neocortex, the area in which a simulation of the external world is rendered
自发恢复（与联想学习有关）：应对世界变化的突发事件并使早期两侧对称动物能够不断学习的技术之一；断裂的联想会被迅速抑制，但实际上并没有被忘记；只要有足够的时间，它们就会重新出现
spontaneous Recovery (in relation to associative learning): one of the techniques to deal with changing contingencies in the world and enable continual learning in early bilaterians; broken associations are rapidly suppressed but not, in fact, unlearned; given enough time, they reemerge
颞上沟（STS）：早期灵长类动物进化而来的感觉新皮层的新区域
superior temporal sulcus (STS): a new region of sensory neocortex that evolved in early primates
突触：神经元之间的连接，化学信号通过该连接传递
synapse: the connection between neurons through which chemical signals are passed
时间信用分配问题：当发生事件或结果时，你会将哪些先前的线索或行动“信用”为预测该事件或结果？当必须在时间上分开的事物之间分配信用时，这是信用分配问题的一个子案例
temporal credit assignment problem: when an event or outcome occurs, what previous cue or action do you give “credit” for being predictive of that event or outcome? This is a subcase of the credit assignment problem when having to assign credit between things separated in time
时间差异学习（TD学习）：无模型强化学习过程，其中人工智能系统（或动物大脑）根据预测的未来奖励（而不是实际奖励）的变化（即“时间差异”）来强化或惩罚行为
temporal difference learning (TD learning): the model-free reinforcement learning process whereby AI systems (or animal brains) reinforce or punish behaviors based on changes (i.e., “temporal differences”) in predicted future rewards (as opposed to actual rewards)
时间差异信号（TD信号）：预测未来奖励的变化；该信号在时间差异学习系统中用作强化/惩罚信号
temporal difference signal (TD signal): the change in predicted future reward; this signal is used as the reinforcement/punishment signal in temporal difference learning systems
颞顶交界处（TPJ）：早期灵长类动物进化而来的感觉新皮层的新区域
temporoparietal junction (TPJ): a new region of sensory neocortex that evolved in early primates
心智理论：推断其他动物意图和知识的能力
theory of mind: the ability to infer another animal’s intent and knowledge
效价：刺激的好坏，在行为上由动物是否会接近或避免刺激来定义
valence: the goodness or badness of a stimulus, behaviorally defined by whether an animal will approach or avoid the stimulus
参考书目
Bibliography
为了节省纸张，完整的参考书目可以在 briefhistoryofintelligence.com 上找到。
To save paper, the full bibliography can be found at briefhistoryofintelligence.com.
在为本书进行研究的这些年里，我阅读了数百本书、论文和期刊，其中绝大多数都在注释部分中被引用。以下作品（按标题字母顺序排列）对本书框架的制定尤为重要。
Over the course of my years of research for this book, there were hundreds of books, papers, and journals that I read—the vast majority of which are cited in the Notes section. The works below (in alphabetical order by title) were particularly important in formulating the framework in this book.
《对齐问题：机器学习与人类价值观》作者：Brian Christian
The Alignment Problem: Machine Learning and Human Values by Brian Christian
行为：人类最佳和最坏状态下的生物学，作者：罗伯特·萨波斯基
Behave: The Biology of Humans at Our Best and Worst by Robert Sapolsky
大脑结构及其起源：行为和心智的发展与进化，作者：Gerald E. Schneider
Brain Structure and Its Origins: In Development and in Evolution of Behavior and the Mind by Gerald E. Schneider
《穿越时空的大脑：脊椎动物的自然历史》（作者：Georg F. Striedter 和 R. Glenn Northcutt）
Brains Through Time: A Natural History of Vertebrates by Georg F. Striedter and R. Glenn Northcutt
埃德蒙·罗尔斯的《大脑皮层》
Cerebral Cortex by Edmund Rolls
《我们自身的深厚历史：四十亿年来我们如何获得意识大脑的故事》作者：约瑟夫·勒杜克斯
The Deep History of Ourselves: The Four-Billion-Year Story of How We Got Conscious Brains by Joseph LeDoux
《深度学习》（作者：Ian Goodfellow、Yoshua Bengio 和 Aaron Courville）
Deep Learning by Ian Goodfellow, Yoshua Bengio, and Aaron Courville
保罗·西塞克 (Paul Cisek) 的《从脊索动物到灵长类动物的行为控制进化》
Evolution of Behavioural Control from Chordates to Primates by Paul Cisek
《语言的进化》作者：W. Tecumseh Fitch
The Evolution of Language by W. Tecumseh Fitch
《记忆系统的进化》（作者：Elisabeth A. Murray、Steven P. Wise 和 Kim S. Graham）
The Evolution of Memory Systems by Elisabeth A. Murray, Steven P. Wise, and Kim S. Graham
《敏感灵魂的进化：学习与意识的起源》（作者：西蒙娜·金斯伯格和伊娃·雅布隆卡）
The Evolution of the Sensitive Soul: Learning and the Origins of Consciousness by Simona Ginsburg and Eva Jablonka
进化神经科学作者：Jon H. Kaas
Evolutionary Neuroscience by Jon H. Kaas
《鱼类认知与行为》（作者：Culum Brown、Kevin Laland 和 Jens Krause）
Fish Cognition and Behavior by Culum Brown, Kevin Laland, and Jens Krause
《从神经元到通过计算神经科学实现的认知》由 Michael A. Arbib 和 James J. Bonaiuto 编辑
From Neuron to Cognition via Computation Neuroscience edited by Michael A. Arbib and James J. Bonaiuto
《差距：人类与其他动物的区别之科学》（作者：托马斯·萨登多夫）
The Gap: The Science of What Separates Us from Other Animals by Thomas Suddendorf
情绪是如何产生的：大脑的秘密生活（作者：丽莎·费尔德曼·巴雷特）
How Emotions Are Made: The Secret Life of the Brain by Lisa Feldman Barrett
《猴子如何看世界：另一个物种的思想》作者：Dorothy L. Cheney 和 Robert M. Seyfarth
How Monkeys See the World: Inside the Mind of Another Species by Dorothy L. Cheney and Robert M. Seyfarth
大脑如何运作：学习和识别的层次和时间模型（作者：Dileep George）
How the Brain Might Work: A Hierarchical and Temporal Model for Learning and Recognition by Dileep George
未来的发明：托马斯·萨登多夫的深思熟虑的自然史
The Invention of Tomorrow: A Natural History of Forethought by Thomas Suddendorf
语言进化(Morten H. Christiansen 和 Simon Kirby 编辑)
Language Evolution edited by Morten H. Christiansen and Simon Kirby
《大脑中的思维：我们如何做出决定以及这些决定为何出错》作者：A. David Redish
The Mind Within the Brain: How We Make Decisions and How Those Decisions Go Wrong by A. David Redish
《前额皮质的神经生物学：解剖学、进化和洞察力的起源》（作者：Richard E. Passingham 和 Steven P. Wise）
The Neurobiology of the Prefrontal Cortex: Anatomy, Evolution, and the Origin of Insight by Richard E. Passingham and Steven P. Wise
神经经济学：决策与大脑（作者：保罗·格里姆彻和恩斯特·费尔）
Neuroeconomics: Decision Making and the Brain by Paul Glimcher and Ernst Fehr
新型高管大脑：复杂世界中的额叶作者：Elkhonon Goldberg
The New Executive Brain: Frontal Lobes in a Complex World by Elkhonon Goldberg
杰夫·霍金斯《论情报》
On Intelligence by Jeff Hawkins
强化学习： Richard S. Sutton 和 Andrew G. Barto 的介绍
Reinforcement Learning: An Introduction by Richard S. Sutton and Andrew G. Barto
通过系统发育改进重新合成行为，作者：Paul Cisek
Resynthesizing Behavior Through Phylogenetic Refinement by Paul Cisek
超级智能：路径、危险、策略作者：尼克·博斯特罗姆
Superintelligence: Paths, Dangers, Strategies by Nick Bostrom
《千个大脑：智力的新理论》作者：杰夫·霍金斯
A Thousand Brains: A New Theory of Intelligence by Jeff Hawkins
为什么黑猩猩不能学习语言，只有人类可以？作者：Herbert S. Terrace
Why Chimpanzees Can’t Learn Language and Only Humans Can by Herbert S. Terrace
为什么只有我们：语言与进化（罗伯特·C·贝里克和诺姆·乔姆斯基著）
Why Only Us: Language and Evolution by Robert C. Berwick and Noam Chomsky
笔记
Notes
介绍
Introduction
“就像家人一样”：《罗西的男朋友》。《杰森一家》，由威廉·汉纳和约瑟夫·巴伯拉创作。第 1 季，第 8 集，1962 年。
“just like one of the family”: “Rosey’s Boyfriend.” The Jetsons, created by William Hanna and Joseph Barbera. Season 1, episode 8, 1962.
“即将实现人类水平的人工智能……”： Cuthbertson，2022 年。
“on the verge of achieving human-level AI.: Cuthbertson, 2022.
超过十亿个连接：突触宽度约为 20 纳米 (Zuber et al., 2005)。一立方毫米内有大约 10 亿个连接 (Faisal et al., 2005)。
over one billion connections: width of a synapse of about 20 nanometers (Zuber et al., 2005). Within a cubic millimeter there are about 1 billion connections (Faisal et al., 2005).
“进入人类大脑”：引自 Hinton 在“多伦多大学计算机科学家因在人工智能领域的开创性工作荣获国际大奖”中的报道。多伦多大学新闻。2017 年 1 月 18 日，https://www.utoronto.ca/news/ut-computer-scientist-takes-international-prize-groundbreaking-work-ai。
“to the human brain”: quote from Hinton reported in “U of T computer scientist takes international prize for groundbreaking work in AI.” U of T News. January 18, 2017, https://www.utoronto.ca/news/u-t-computer-scientist-takes-international-prize-groundbreaking-work-ai.
“运动和其他功能”： MacLean，1990 年。
“motor, and other functions”: MacLean, 1990.
以及它的工作原理： Cesario 等人，2020 年，对 MacLean 三位一体大脑模型的当前观点进行了很好的概述。尽管公平地说，在我看来，他的三位一体大脑模型的大多数问题都在于其广受欢迎的成功。如果有人真的读过 MacLean 的作品，他就会很容易地承认他的框架存在许多挑战。
and how it works: Cesario et al., 2020, provides a good overview of the current view of MacLean’s triune brain model. Although, to be fair to MacLean, it seems to me that most of the issues with his triune brain model are in its popular success. If one actually reads MacLean’s work, he readily acknowledges many of the challenges with his framework.
“这就是人类级别的人工智能（AHI）”： Yann LeCun（@ylecun）于 2019 年 12 月 9 日发布了这条推文。
“so on to Artificial Human–level Intelligence (AHI)”: Yann LeCun (@ylecun) tweeted this on December 9, 2019.
比人类更快： Healy 等人，2013 年。
faster than a human: Healy et al., 2013.
第一章：大脑之前的世界
Chapter 1: The World Before Brains
不起眼的热液喷口：有关热液喷口理论和生命时间的评论，请参阅 Bell 等人，2015 年；Dodd 等人，2017 年；Martin 等人，2008 年；McKeegan 等人，2007 年。
an unremarkable hydrothermal vent: For reviews of hydrothermal-vent theory and the timing of life, see Bell et al., 2015; Dodd et al., 2017; Martin et al., 2008; McKeegan et al., 2007.
它复制了自身：有关 RNA 世界的评论以及 RNA 最初可以在没有蛋白质的情况下复制自身的证据，请参阅 Neveu 等人，2013 年。
it duplicated itself: For a review of RNA world and the evidence that RNA could originally duplicate itself without proteins, see Neveu et al., 2013.
现代船舶：细菌鞭毛由质子驱动，通过旋转马达运转。参见 Lowe 等人，1987 年；Silverman 和 Simon，1974 年。
of a modern boat: Bacterial flagellum is proton-driven and works via rotary motor that turns. See Lowe et al., 1987; Silverman and Simon, 1974.
合成、脂质和碳水化合物：有关 LUCA 具有 DNA 的证据，请参阅 Hassenkam 等人，2017 年。有关 LUCA 正在进行蛋白质合成的证据，请参阅 Noller，2012 年。
synthesis, lipids, and carbohydrates: For evidence LUCA had DNA, see Hassenkam et al., 2017. For evidence LUCA was performing protein synthesis, see Noller, 2012.
为这些流程提供资金： JLE Wimmer 等人，2021 年。
finance these many processes: J. L. E. Wimmer et al., 2021.
蓝藻：请注意，科学家不再喜欢蓝藻这个术语，因为藻类这个词专门用于指一种单细胞植物。
blue-green algae: Note scientists don’t like the term blue-green algae anymore since the word algae is reserved for a type of single-celled plant.
图 1.1：图片来自 https://www.scienceimage.csiro.au/image/4203。CC BY 3.0 许可证，由 Willem van Aken 于 1993 年 3 月 18 日拍摄。
Figure 1.1: Figure from https://www.scienceimage.csiro.au/image/4203. CC BY 3.0 license Photograph by Willem van Aken on March 18, 1993.
转化为细胞能量：通过光合作用产生氧气的能力可能最早出现在蓝藻的祖先中；参见 KL French 等人，2015 年。
converted into cellular energy: The ability to generate oxygen via photosynthesis likely first appeared in the ancestors of cyanobacteria; see K. L. French et al., 2015.
并不断复制： Cardona et al., 2015； Schirrmeister 等人，2013。
and endlessly reproducing: Cardona et al., 2015; Schirrmeister et al., 2013.
产生污染性废气：请注意，早期生命可能使用效率较低且更原始的光合作用，这种光合作用产生的能量较少，并且不会产生氧气作为废气。参见 Raymond 和 Segrè，2006 年。
produced a pollutive exhaust: Note that earlier life may have used a less efficient and more primitive version of photosynthesis, one that produced less energy and did not produce oxygen as exhaust. See Raymond and Segrè, 2006.
氧气水平猛增：TW Lyons 等人，2014 年。
oxygen levels skyrocketed: T. W. Lyons et al., 2014.
氧气大屠杀： Margulis 和 Sagan，1997 年。
the Oxygen Holocaust: Margulis and Sagan, 1997.
地球历史上的灭绝事件：大氧化事件发生在公元前 24 亿年；参见 Anbar 等人，2007 年。有关它消灭了地球上许多物种的证据，参见 Hodgskiss 等人，2019 年。
extinction events in Earth’s history: The Great Oxygenation Event occurred at 2.4B BCE; see Anbar et al., 2007. For evidence it killed off many species on Earth, see Hodgskiss et al., 2019.
二氧化碳作为废气：从技术上讲，进化出来的是有氧呼吸，即利用氧气的呼吸方式。有关蓝藻产氧光合作用后进化出有氧呼吸的证据，请参阅 Soo et al., 2017。
carbon dioxide as exhaust: Technically, what evolved is aerobic respiration, the version of respiration that uses oxygen. For evidence that aerobic respiration evolved after oxygenetic photosynthesis in cyanobacteria, see Soo et al., 2017.
基于氧的方法（有氧呼吸）： O'Leary 和 Plaxton，2016 年。
oxygen-based approach (aerobic respiration): O’Leary and Plaxton, 2016.
内部复杂性大大增强：真核生物的复杂性可以在显微镜下看到；真核生物这一术语源于 20 世纪的观察，即真核生物的后代都有好的（ eu）核（ karyon），而细菌和类细菌生命没有这样的内部结构，因此被称为原核生物：在（ pro）核（ karyon）之前。
much more internal complexity: The complexity of eukaryotes can be seen under a microscope; the term eukaryote comes from the twentieth-century observation that the descendants of eukaryotes all had good (eu) kernels (karyon), while bacteria and bacteria-like life had no such inner structures and hence were called prokaryotes: before (pro) kernels (karyon).
微生物杀人机器：有关真核生物是第一个吞噬并在内部消化生命作为食物的细胞的证据，请参阅 Cavalier-Smith，2009。
microbial killing machines yet: For evidence eukaryotes were the first cells to engulf and internally digest life for food, see Cavalier-Smith, 2009.
生命之树：真核生物的出现时间约为 20 亿年前；参见 Knoll 等人，2006 年。
The tree of life: Timing emergence of eukaryotes to around 2B years ago; see Knoll et al., 2006.
蘑菇状真菌开始生长： Bengtson 等人，2017 年。
mushroom-like fungi began growing: Bengtson et al., 2017.
图 1.5：Reichert，1990 年的插图。已获许可使用。
Figure 1.5: Illustration from Reichert, 1990. Used with permission.
所有神经元都来自哪里：可能有一个例外：栉水母可能有独立进化的神经元。
whom all neurons descend: There may be a single exception to this: comb jellies might have independently evolved neurons.
地球上的生命比动物更多： Bar-On 等人，2018 年。
on Earth than animals: Bar-On et al., 2018.
与水母胚胎一样多： Technau，2020年。
as much as jellyfish embryos: Technau, 2020.
神经元使能的动物祖先：从今以后，当我说动物时，我指的是真后生动物，即“真正的”后生动物。
neuron-enabled animal ancestor: Henceforth when I say animals, I’m referring to Eumetazoa, the “true” metazoans.
具有神经元的原肠胚状生物： Arendt 等人，2016 年。
gastrula-shaped creature with neurons: Arendt et al., 2016.
与今天的珊瑚相似： Penny 等人，2014 年；Wan 等人，2016 年。
similar to today’s corals: Penny et al., 2014; Wan et al., 2016.
错误的开始和错误的转折：有关神经系统历史发现的回顾，请参阅 McDonald，2004。
false starts and wrong turns: For a review of the historical discovery of nervous systems, see McDonald, 2004.
不是以太而是电：Piccolino，1997；Schuetze，1983。
not ether but electricity: Piccolino, 1997; Schuetze, 1983.
产生自己的信号： O'Brien，2006。
generate their own signals: O’Brien, 2006.
阿德里安·诺贝尔奖：加森，2015 年；皮尔斯，2018 年。
Adrian the Nobel Prize: Garson, 2015; Pearce, 2018.
尖峰或动作电位：动作电位的发现实际上发生得更为缓慢，有人认为它早在 1848 年就发生了。参见 du Bois-Reymond，E. 1848。
spikes or action potentials: The discovery of action potentials actually happened more gradually, some suggesting it happened as early as 1848. See du Bois-Reymond, E. 1848.
神经元的信号： Garson，2003。
the signals of neurons: Garson, 2003.
尖峰本身：与生物学中的大多数事物一样，也有例外：大脑的某些区域似乎使用其他编码策略，例如时间编码。
of the spike itself: As with most everything in biology, there are exceptions: some areas of brains seem to use other coding strategies, such as temporal coding.
从水母到人类：有关水螅的速率编码，请参阅 Tzouanas 等人，2021 年。有关秀丽隐杆线虫的速率编码，请参阅 Q. Liu 等人，2018 年；O'Hagan 等人，2005 年；Suzuki 等人，2003 年。
from jellyfish to humans: For rate coding in hydra, see Tzouanas et al., 2021. For rate coding in C. elegans, see Q. Liu et al., 2018; O’Hagan et al., 2005; Suzuki et al., 2003.
他们的发放率： JT Pearson 和 D. Kerschensteiner，2015 年。
in their firing rate: J. T. Pearson and D. Kerschensteiner, 2015.
浓度在其放电率中的作用： Parabucki 等人，2019 年。
concentration in their firing rate: Parabucki et al., 2019.
肌肉力量：这甚至在秀丽隐杆线虫中也得到证实；参见 S. Gao 和 M. Zhen，2011 年。
force of the muscles: This has been shown even in C. elegans; see S. Gao and M. Zhen, 2011.
月光下的一页： MacEvoy，B. 2015
a page in moonlight: MacEvoy, B. 2015
图 1.10：图来自 B. MacEvoy，2015 年。已获许可使用（个人通信）。
Figure 1.10: Figure from B. MacEvoy, 2015. Used with permission (personal correspondence).
每秒五百个尖峰： Wang 等人，2016 年。
five hundred spikes per second: Wang et al., 2016.
约翰·埃克尔斯 (John Eccles) 等：埃克尔斯发现了抑制；戴尔发现了化学神经传递 (Todman, 2008)；谢灵顿发现了突触 (RE Brown 等人，2021)。
John Eccles, and others: Eccles discovered inhibition; Dale discovered chemical neurotransmission (Todman, 2008); Sherrington discovered synapses (R. E. Brown et al., 2021).
吞咽反射起作用：在水螅中发现了通过突触抑制而产生侧抑制的证据；参见 Kass-Simon，1988 年。但一些人认为，刺胞动物中不存在突触抑制（Meech 和 Mackie，2007 年）。
swallow reflexes to work: Evidence of lateral inhibition through synaptic inhibition has been found in the hydra; see Kass-Simon, 1988. But some have argued that synaptic inhibition is absent in Cnidaria (Meech and Mackie, 2007).
另一个必须放松：有关海葵口部肌肉的详细信息，请参阅 Bocharova 和 Kozevich（2011 年）。
and another must relax: See Bocharova and Kozevich, 2011, for details on mouth muscles of sea anemones.
第二章 善与恶的诞生
Chapter 2: The Birth of Good and Bad
一粒米：有关早期两侧对称动物的化石证据，请参阅 Z. Chen 等人，2013 年、2018 年；Evans 等人，2020 年。
a grain of rice: For fossil evidence of early bilaterians, see Z. Chen et al., 2013, 2018; Evans et al., 2020.
几乎肯定非常简单：有关早期两侧对称动物的详细评论，请参阅 Malakhov，2010。
almost definitely very simple: For a great review of early bilaterians, see Malakhov, 2010.
人类的 850 亿个：更准确地说，秀丽隐杆线虫整个身体有 302 个神经元，而人类的大脑中有 850 亿个神经元；人类神经系统中除了大脑之外还有其他神经元。
a human’s 85 billion: To be more precise, C. elegans has 302 neurons in its entire body, while a human has 85 billion neurons in his or her brain; there are other neurons in the human nervous system outside of the brain.
找到食物：从今以后，每当我使用线虫这个术语时，我指的是特定种类的线虫，即秀丽隐杆线虫。
finds the food: Henceforth whenever I use the term nematode, I am referring to the specific species of nematode Caenorhabditis elegans.
关于食物：有关扁虫的例子，请参阅 Pearl, 1903。有关线虫的例子，请参阅 Bargmann et al., 1993；Ward, 1973。
in on the food: For an example with flatworms, see Pearl, 1903. For nematodes, see Bargmann et al., 1993; Ward, 1973.
身体可能会受伤：有关秀丽隐杆线虫此类行为的详细综述，请参阅 Hobert, 2003。有关秀丽隐杆线虫的热梯度行为请参阅 Cassata et al., 2000；Hedgecock and Russell, 1975；L. Luo et al., 2014。
bodies might get wounded: For a good review of these types of behaviors in C. elegans, see Hobert, 2003. For thermal-gradient behavior in C. elegans, see Cassata et al., 2000; Hedgecock and Russell, 1975; L. Luo et al., 2014.
“从中获得”： Brooks，1991年。
“be gained from it”: Brooks, 1991.
“圣杯就在他们手中”：同上。
“grail within their grasp”: Ibid.
超过四千万台：历史 | iRobot。（nd）。2023 年 3 月 5 日检索自 https://about.irobot.com/History。
over forty million units: History | iRobot. (n.d.). Retrieved March 5, 2023, from https://about.irobot.com/History.
或识别物体：请注意，Roomba 的未来版本确实增加了允许它学习房屋地图的功能。
or to recognize objects: Note that future versions of the Roomba did add features that allowed it to learn maps of the house.
图 2.8：拉里·D·摩尔 (Larry D. Moore) 于 2006 年拍摄的照片。图片根据 CC-BY 许可在维基百科 https://en.wikipedia.org/wiki/Roomba 上发布。
Figure 2.8: Photograph by Larry D. Moore in 2006. Picture published on Wikipedia at https://en.wikipedia.org/wiki/Roomba under a CC-BY license.
到不同的下游神经元： Garrity 等人，2010 年。
to different downstream neurons: Garrity et al., 2010.
当温度太冷时，它们会变冷： L. Luo 等人，2014 年。有关更多详细信息，请参阅 Garrity 等人，2010 年的精彩评论。当温度高于阈值时，“AFD”神经元仅对“太热”的温度做出反应（参见 Goodman 和 Sengupta，2018 年）。
cold when they’re too cold: L. Luo et al., 2014. For additional details see the great review by Garrity et al., 2010. The “AFD” neuron only responds to “too hot” temperatures when temperature is above a threshold (see Goodman and Sengupta, 2018).
跨越障碍： Hobert，2003；Ishihara 等人，2002。
to cross the barrier: Hobert, 2003; Ishihara et al., 2002.
跨越不同的感官模式： Inoue 等人，2015 年。
across different sensory modalities: Inoue et al., 2015.
铜屏障：这种相互抑制的具体电路更复杂，但原理相似。秀丽隐杆线虫中有一个名为 AWC 的感觉神经元，它会被正价气味所激发。秀丽隐杆线虫中有四个下游神经元从感觉神经元获取输入；这些下游神经元分别名为 AIZ、AIB、AIY 和 AIA。AIZ 和 AIB 促进转弯，而 AIY 和 AIA 促进向前运动（Garrity 等人，2010 年讨论）。这些下游神经元之间存在相互抑制：AIY 抑制 AIB（Chalasani 等人，2007 年），AIY 抑制 AIZ（Z. Li 等人，2014 年），AIA 抑制 AIB（Wakabayashi 等人，2004 年）。一些相互抑制发生在更下游；例如，通过将 AIY 的抑制输出与 AIB 的兴奋输入整合到另一个神经元 RIB 上，从而促进转弯本身（Garrity 等人，2010；JM Gray 等人，2005）。电路很混乱，但效果是一样的；前进投票和转弯投票之间存在相互抑制。
copper barrier: The exact circuit for this mutual inhibition is more complex but similar in principle. In C. elegans there is a sensory neuron named AWC, which gets excited by positive valence smells. There are four downstream neurons in C. elegans that get input from sensory neurons; these downstream neurons are named AIZ, AIB, AIY, and AIA. AIZ and AIB promote turning, while AIY and AIA promote forward movement (discussed in Garrity et al., 2010). There is mutual inhibition among these downstream neurons: AIY inhibits AIB (Chalasani et al., 2007), AIY inhibits AIZ (Z. Li et al., 2014), and AIA inhibits AIB (Wakabayashi et al., 2004). Some of the mutual inhibition occurs further downstream; for example, by integrating inhibitory output from AIY with excitatory input from AIB on another neuron RIB, which promotes turns itself (Garrity et al., 2010; J. M. Gray et al., 2005). The circuit is messy, but the effect is the same; there is mutual inhibition between votes for moving forward and votes for turning.
饥饿的线虫是： Davis 等人，2017 年；Lau 等人，2017 年。
hungry a nematode is: Davis et al., 2017; Lau et al., 2017.
他们有多饿： Rengarajan 等人，2019 年。
how hungry they are: Rengarajan et al., 2019.
健康的能量： Davis 等人，2017 年。
healthy amount of energy: Davis et al., 2017.
第三章 情绪的起源
Chapter 3: The Origin of Emotion
每个单词单独使用： Jackson 等人，2019 年。
separate words for each: Jackson et al., 2019.
情绪的两种属性： Barrett 和 Russell，1999 年；Russell，2003 年。
two attributes of emotions: Barrett and Russell, 1999; Russell, 2003.
肾上腺素和血压： Heilman，1997 年；Lang、Bradley 和 Cuthbert，1997 年。
adrenaline, and blood pressure: Heilman, 1997; Lang, Bradley, and Cuthbert, 1997.
特定大脑区域： Gerber 等人，2008 年。
of specific brain regions: Gerber et al., 2008.
价态和唤醒： Jackson 等人，2019 年；Wierzbicka，1992 年。
of valence and arousal: Jackson et al., 2019; Wierzbicka, 1992.
（例如哭泣和微笑）： Bridges，1932 年；Graf，2015 年；Huffman，1997 年；Oster，2012 年；Saarni 等人，2006 年。
(e.g., crying and smiling): Bridges, 1932; Graf, 2015; Huffman, 1997; Oster, 2012; Saarni et al., 2006.
食物没了： Hills 等人，2004 年；D. Jones 和 Candido，1999 年；Z. Liu 等人，2018 年。
the food is gone: Hills et al., 2004; D. Jones and Candido, 1999; Z. Liu et al., 2018.
大脑中的多巴胺： Chase 等人，2004 年。
the brain with dopamine: Chase et al., 2004.
剥削状态：如果破坏这些多巴胺神经元，对食物的剥削行为就会消失（Sawin 等人，2000 年）。Hills 等人（2004 年）证明了这一点，并且还表明，如果保持这些神经元完好无损，但阻止神经元之间的多巴胺信号传递，剥削行为同样会消失，如果将多巴胺注入秀丽隐杆线虫的大脑，它会立即表现出剥削行为（减慢并增加转动频率，就像它检测到了食物一样）。如果在多巴胺神经元被破坏后向秀丽隐杆线虫体内注入多巴胺
the state of exploitation: If you destroy these dopamine neurons, exploitation behavior in response to food goes away (Sawin et al., 2000). Hills et al., 2004, shows this and also shows that if you leave those neurons intact but prevent dopamine signaling between neurons, exploitation behavior similarly goes away, and if you inject dopamine into the brain of C. elegans, it immediately shows exploitation (slowing down and increasing turning frequency as if it had detected food). And if you inject dopamine into C. elegans even after dopamine neurons have been destroyed, exploitation behavior returns.
多巴胺神经元激发后，多巴胺会长时间停留在细胞外液中，从而产生这种持续状态。多巴胺通过直接调节整套神经元的反应来实现这一点。例如，多巴胺调节特定运动神经元的反应（Chase 等人，2004 年）和转向神经元的反应（Hills 等人，2004 年），并直接调节价神经元（Sanyal 等人，2004 年）。所有这些精心策划的调节的结果是，你会得到一种新的情感开发状态，其中蠕虫移动得更慢，转弯更频繁。
Dopamine generates this persistent state by remaining in the extracellular fluid long after the dopamine neurons fired. Dopamine accomplishes this by directly modulating the responses of a whole suite of neurons. For example, dopamine modulates the responses of specific motor neurons (Chase et al., 2004) and the responses of the steering neurons (Hills et al., 2004), and it modulates valence neurons directly (Sanyal et al., 2004). The consequence of all this orchestrated modulation is that you get a new affective state of exploitation in which worms move more slowly and turn more frequently.
喉咙里的食物： Rhoades 等人，2019 年。
food in their throats: Rhoades et al., 2019.
释放，就会引发饱腹感：有关血清素因检测到胃中的食物而释放的证据，请参阅 Gürel 等人，2012 年。
released, it triggers satiety: For evidence serotonin is released by the detection of food in the stomach, see Gürel et al., 2012.
如果破坏两个与食物相关的血清素神经元，饥饿的线虫在遇到食物时就不会再放慢速度；参见 Sawin 等人，2000 年。如果阻止血清素信号传导，线虫在吃饱时休息的时间几乎不会比饥饿时多（Churgin 等人，2017 年）。如果没有血清素信号传导，线虫在饥饿时会花费更多时间进行逃跑/游荡行为，就好像它们在吃饱后需要更长的时间才能停止寻找食物（Churgin 等人，2017 年；Flavell 等人，2013 年）。对此的一种解释是，没有血清素，线虫很难获得满足。如果将血清素注入秀丽隐杆线虫的大脑，它在饥饿时寻找食物的时间就会少得多（Churgin 等人，2017 年）。血清素还能增加产卵量（Waggoner 等人，1998 年）、交配行为（Loer 和 Kenyon，1993 年；Ségalat 等人，1995 年）以及咽部泵动（相当于吞咽）（Ségalat 等人，1995 年）。
If you destroy the two food-related serotonin neurons, hungry worms no longer additionally slow down when they encounter food; see Sawin et al., 2000. If you prevent serotonin signaling, worms spend barely any additional time resting when full than when hungry (Churgin et al., 2017). Without serotonin signaling, worms spend far more time in escape/roaming behavior when hungry, as if it takes much longer for them to stop looking for food once they get full (Churgin et al., 2017; Flavell et al., 2013). One interpretation of this is that without serotonin, worms struggle to get satisfied. If you inject serotonin into the brain of C. elegans, it spends much less time moving around looking for food when hungry (Churgin et al., 2017). Serotonin also increases egg laying (Waggoner et al., 1998), mating behavior (Loer and Kenyon, 1993; Ségalat et al., 1995), and pharyngeal pumping, the equivalent of swallowing (Ségalat et al., 1995).
追求奖励（饱腹感）：血清素在两侧对称动物中的作用相似（Gillette，2006；Tierney，2020）。食物在口中时会释放血清素，并引发软体动物（Kabotyanski 等人，2000；Yeoman 等人，1994）、线虫（Hobson 等人，2006；Szø 等人，2000）和环节动物（Groome 等人，1995）的吞咽。在脊椎动物中，即使是预期的正价刺激体验也会触发血清素释放（Z. Liu 等人，2020；Zhong 等人，2017）。血清素对攻击性的作用似乎也得到了保留，因为血清素会降低大鼠（Nikulina 等人，1991 年）、鸡（Dennis 等人，2008 年）和甲壳类动物（Kravitz，2000 年）的攻击性。血清素始终在饱腹感中发挥作用，尽管存在一些差异。血清素会引起大鼠（Blundell 和 Leshem，1975 年；Grinker 等人，1980 年）、非人类灵长类动物 (Foltin and Moran, 1989)、人类 (McGuirk et al., 1991; Rogers and Blundell, 1979)、苍蝇 (Long et al., 1986)、蟑螂 (Haselton et al., 2009)、蚂蚁 (Falibene et al., 2012)、蜜蜂 (AS French et al., 2014) 和蚊子 (Ling and Raikhel, 2018)。然而，在环节动物和软体动物中，情况似乎有所不同；血清素似乎会引起饥饿并增加环节动物 (Lent et al., 1991) 和软体动物 (Hatcher et al., 2008; Yeoman et al., 1994) 的进食量，并降低两者的进食阈值 (Palovcik et al., 1982)。
pursuit of rewards (satiation): The role of serotonin is similar across Bilateria (Gillette, 2006; Tierney, 2020). Serotonin is released when food is in the mouth and triggers swallowing in mollusks (Kabotyanski et al., 2000; Yeoman et al., 1994), nematodes (Hobson et al., 2006; Szø et al., 2000), and annelids (Groome et al., 1995). In vertebrates, the experience of a positive-valanced stimulus, even if expected, triggers serotonin release (Z. Liu et al., 2020; Zhong et al., 2017). The role of serotonin on aggression seems to be conserved as well, as serotonin decreases aggressiveness in rats (Nikulina et al., 1991), chickens (Dennis et al., 2008), and crustaceans (Kravitz, 2000). Serotonin consistently plays a role in satiation, although there are some differences. Serotonin induces satiety and reduces feeding in rats (Blundell and Leshem, 1975; Grinker et al., 1980), nonhuman primates (Foltin and Moran, 1989), humans (McGuirk et al., 1991; Rogers and Blundell, 1979), flies (Long et al., 1986), cockroaches (Haselton et al., 2009), ants (Falibene et al., 2012), honeybees (A. S. French et al., 2014), and mosquitoes (Ling and Raikhel, 2018). However, in annelids and mollusks this seems to be different; serotonin seems to induce hunger and increase feeding in annelids (Lent et al., 1991) and mollusks (Hatcher et al., 2008; Yeoman et al., 1994) and lowers feeding thresholds across both (Palovcik et al., 1982).
大脑释放血清素：有关血清素的详细评论，请参阅 Z. Liu 等人，2020 年的文章。
brain releases serotonin: For a great review of serotonin, see Z. Liu et al., 2020.
与他们见面的任何人： Musselman 等人，2012 年。
with whomever they see: Musselman et al., 2012.
愿意延迟满足：有关提高血清素水平会减少进食的证据，请参阅 Sharma 和 Sharma，2012 年。有关提高血清素水平会增加延迟满足意愿的证据，请参阅 Linnoila 等人，1983 年。
willing to delay gratification: For evidence that raising serotonin decreases eating, see Sharma and Sharma, 2012. For evidence that raising serotonin increases willingness to delay gratification, see Linnoila et al., 1983.
减弱价神经元的反应：有关关闭多巴胺反应的信息，请参阅 Valencia-Torres 等人，2017 年。有关减弱价反应的信息，请参阅 Lorrain 等人，1999 年。
dulling the responses of valence neurons: For turning off dopamine responses, see Valencia-Torres et al., 2017. For dulling valence responses, see Lorrain et al., 1999.
图 3.5： Kent Berridge 的图片（私人信件）。已获许可使用。
Figure 3.5: Images from Kent Berridge (personal correspondence). Used with permission.
食物而饿死： 1998 年 Berridge 和 Robinson 评论。
food and starve to death: Reviewed in Berridge and Robinson, 1998.
每小时次数： Heath，1963 年。
of times an hour: Heath, 1963.
“达到终点”：诚然，希思认为他发现的证据表明隔膜刺激是令人愉悦的，他确实声称患者似乎感觉“良好”（希思，1963 年）。但其他实验者做了同样的实验，发现“刺激过程中没有‘喜欢’效应，与希思报告的结果相反”（Schlaepfer 等人，2008 年）。
“reach the end point”: Admittedly, Heath believed that the evidence he found showed that septal stimulation was pleasurable, and he did claim that patients seemed to feel “good” (Heath, 1963). But other experimenters did the same experiments and found “there were no ‘liking’ effects during stimulation, in contrast to findings reported by Heath” (Schlaepfer et al., 2008).
喜欢和不喜欢的反应： Treit 和 Berridge，1990 年。
both liking and disliking reactions: Treit and Berridge, 1990.
每年有 1500 万人自杀： Morgan 等人，2018 年。
15 million annual suicide attempts: Morgan et al., 2018.
并参与生活：抑郁症。世界卫生组织，2021 年 9 月 13 日。于 2023 年 3 月 5 日访问，网址为 https://www.who.int/news-room/fact-sheets/detail/depression。
and engage in life: Depression. World Health Organization, 13 September 2021. Accessed on March 5, 2023, at https://www.who.int/news-room/fact-sheets/detail/depression.
睡眠、生殖和消化：去甲肾上腺素对许多（如果不是所有的话）脊椎动物（包括鱼类）具有高度的兴奋性（Singh 等人，2015 年）。章鱼胺（一种相关化合物）同样会增加各种原口动物的兴奋性，例如环节动物（Crisp 等人，2010 年）、节肢动物（Crocker 和 Sehgal，2008 年；Florey 和 Rathmayer，1978 年）和线虫（Churgin 等人，2017 年）。去甲肾上腺素会增加许多脊椎动物的攻击性，包括小鼠（Marino 等人，2005 年）。章鱼胺同样会增加苍蝇的攻击性（C. Zhou 等人，2008 年）。脊椎动物在饥饿时会释放去甲肾上腺素（PJ Wellman，2000 年）。章鱼胺在饥饿状态下会释放，增加节肢动物（Long and Murdock，1982）、软体动物（Vehovszky and Elliott，2002）和线虫（Guo et al.，2015；Suo et al.，2006）的食物消耗。章鱼胺抑制节肢动物的求偶条件作用（C. Zhou et al.，2012），并抑制节肢动物（Sombati and Hoyle，1984）和线虫（Alkema et al.，2005；Guo et al.，2015；Horvitz et al.，1982）的产卵。
sleep, reproduction, and digestion: Norepinephrine is highly arousing across many, if not all, vertebrates, including fish (Singh et al., 2015). Octopamine (a related compound) similarly increases arousal in diverse protostomes such as annelids (Crisp et al., 2010), arthropods (Crocker and Sehgal, 2008; Florey and Rathmayer, 1978), and nematodes (Churgin et al., 2017). Norepinephrine increases aggression across many vertebrates, including mice (Marino et al., 2005). Octopamine similarly increases aggression in flies (C. Zhou et al., 2008). Norepinephrine is released by starvation in vertebrates (P. J. Wellman, 2000). Octopamine is released by starvation and increases food consumption in arthropods (Long and Murdock, 1982), mollusks (Vehovszky and Elliott, 2002), and nematodes (Guo et al., 2015; Suo et al., 2006). Octopamine suppresses courtship conditioning in arthropods (C. Zhou et al., 2012), and suppresses egg laying in arthropods (Sombati and Hoyle, 1984) and nematodes (Alkema et al., 2005; Guo et al., 2015; Horvitz et al., 1982).
章鱼胺/去甲肾上腺素和多巴胺的具体价态可能在节肢动物中发生了翻转（Barron 等人，2010 年综述）：章鱼胺介导蟋蟀（Mizunami 等人，2009 年；Mizunami 和 Matsumoto，2017 年）、蜜蜂（Farooqui 等人，2003 年；Hammer，1993 年）、苍蝇（Schwaerzel 等人，2003 年）和螃蟹（Kaczer 和 Maldonado，2009 年）中的食欲信号。多巴胺可能介导蟋蟀（Mizunami 等人，2009 年；Mizunami 和 Matsumoto，2017 年）和苍蝇（Schwaerzel 等人，2003 年）中的厌恶信号。然而，故事并不那么清楚；奖励会在黄蜂中产生多巴胺依赖的积极情感状态 (Perry et al., 2016)。此外，章鱼胺的缺乏会损害节肢动物苍蝇的厌恶学习 (Mosca, 2017)，而不同亚群的章鱼胺神经元似乎会触发节肢动物苍蝇的趋近或厌恶 (Claßen and Scholz, 2018)。
The specific valence of octopamine/norepinephrine and dopamine may have been flipped in arthropods (reviewed in Barron et al., 2010): Octopamine mediates appetitive signals in crickets (Mizunami et al., 2009; Mizunami and Matsumoto, 2017), honeybees (Farooqui et al., 2003; Hammer, 1993), flies (Schwaerzel et al., 2003), and crabs (Kaczer and Maldonado, 2009). Dopamine may instead mediate aversive signals in crickets (Mizunami et al., 2009; Mizunami and Matsumoto, 2017) and flies (Schwaerzel et al., 2003). However, the story is not so clear; rewards create dopamine-dependent positive-affective states in bumblebees (Perry et al., 2016). Further, lack of octopamine impairs aversive learning in arthropod flies (Mosca, 2017), and different subsets of octopamine neurons seem to trigger either approach or aversion in arthropod flies (Claßen and Scholz, 2018).
不可否认，有许多唤起化学物质具有略微不同的效果（D.Chen 等人，2016 年）。但这仍然是一个有指导意义的“初次尝试”，而且值得注意的是，单个神经调质与线虫特定情感状态的对应关系非常良好。如果阻断去甲肾上腺素，即使暴露于有害刺激，线虫在逃避行为上所花的时间也会大大减少，而固定时间则会大大增加（Churgin 等人，2017 年）。线虫失去了进入“我必须离开这里去找食物！”模式的能力。去甲肾上腺素实现这一目标的方式与任何其他神经调质相同——通过持续调节控制运动和转动的各种神经元（Rengarajan 等人，2019 年）。与其他神经调质一样，去甲肾上腺素也会调节价神经元。没有去甲肾上腺素，线虫在饥饿时就无法将其行为从避免二氧化碳转变为吸引（Rengarajan 等人，2019 年）。
Admittedly, there are many arousing chemicals with slightly different effects (D. Chen et al., 2016). But this is still an instructive “first pass” and it is remarkable how well individual neuromodulators map to specific affective states in nematodes. If you block norepinephrine, worms spend dramatically less time in their escape behavioral repertoire and far more time immobilized, even if exposed to noxious stimuli (Churgin et al., 2017). Worms lose the ability to get into the mode of “I have to get out of here and find food!” Norepinephrine accomplishes this the same way any other neuromodulator does—by persistently modulating various neurons that control movement and turning (Rengarajan et al., 2019). Like other neuromodulators, norepinephrine also modulates valence neurons. Without norepinephrine, worms fail to shift their behavior from CO2 avoidance to attraction when starved (Rengarajan et al., 2019).
休息并感到满足： Churgin 等人，2017 年；Rex 等人，2004 年；Suo 等人，2006 年。
rest and be content: Churgin et al., 2017; Rex et al., 2004; Suo et al., 2006.
肌肉的能量资源：具体来说，是一系列与肾上腺素相关的化合物（去甲肾上腺素、章鱼胺和肾上腺素）。
energetic resources to muscles: Specifically, the suite of adrenaline related compounds (norepinephrine, octopamine, and epinephrine).
不能无限期地持续下去：请参阅 Sapolsky 等人（2000 年）对压力反应的全面概述。这个类比是受 Sapolsky（2004 年）启发的。
cannot go on indefinitely: See Sapolsky et al., 2000, for a great overview of the stress response. This analogy was inspired by Sapolsky, 2004.
应对压力源： Park 等人，2020 年。
in response to stressors: Park et al., 2020.
受到急性应激源抑制： Staub 等人，2012 年。
inhibited by acute stressors: Staub et al., 2012.
长时间进食： Cheong 等人，2015 年。
prolonged bouts of feeding: Cheong et al., 2015.
抑制疼痛反应： Mills 等，2016；涅托-费尔南德斯等人，2009；普赖尔等人，2007。
inhibited pain responses: Mills et al., 2016; Nieto-Fernandez et al., 2009; Pryor et al., 2007.
并抑制生殖行为： Ow 和 Hall，2020 年；Seidel 和 Kimble，2011 年。
and inhibited reproductive behavior: Ow and Hall, 2020; Seidel and Kimble, 2011.
他们通常饥饿的同龄人： You 等人，2008 年。
their normally hungry peers: You et al., 2008.
暂停身体机能： Nath 等人，2016 年。
pause bodily functions: Nath et al., 2016.
他们放弃了： AJ Hill 等人，2014 年。
they give up: A. J. Hill et al., 2014.
慢性应激状态：在暴露于高温等应激源后，蠕虫会表现出一段有助于生存的静止期。参见 Fry 等人，2016 年；AJ Hill 等人，2014 年；Konietzka 等人，2019 年；van Buskirk 和 Sternberg，2007 年。饥饿也是如此；参见 Park 等人，2020 年；Skora 等人，2018 年。
state of chronic stress: Following exposure to a stressor like heat, the worms exhibit a period of quiescence that aids survival. See Fry et al., 2016; A. J. Hill et al., 2014; Konietzka et al., 2019; van Buskirk and Sternberg, 2007. The same is true of starvation; see Park et al., 2020; Skora et al., 2018.
反应、食欲和生殖： Adamo 和 Baker，2011 年。
response, appetite, and reproduction: Adamo and Baker, 2011.
压力开始激活血清素：在经过基因编辑的秀丽隐杆线虫中，胰岛素信号减弱（从而引发类似抑郁的状态），你会看到这样的情况：如果血清素被阻断，这些看似永久抑郁的线虫就会不再抑郁（Dagenhardt 等人，2017 年）。它们会重新四处走动并对食物提示做出反应。有趣的是，这种血清素抑郁症治疗方法似乎对从秀丽隐杆线虫到人类都有效。治疗人类抑郁症的主要药物是选择性血清素再摄取抑制剂 (SSRI)，例如百忧解，有证据表明它可以降低大脑中的血清素水平。这里很复杂。起初，SSRI 实际上通过阻断突触对血清素的再摄取来增加大脑中的血清素水平。但数周后，这种阻断会改变血清素神经元的反应，使它们减少信号传递，因此最终效果是降低血清素水平。这就是为什么 SSRI 一开始会使抑郁症恶化，但在两到六周后，人们会开始感觉好些。当然，关于这个故事是否完全正确仍然存在争议。
stress starts activating serotonin: In genetically edited C. elegans worms with reduced insulin signaling (which triggers a depression-like state), this is what you see: if serotonin is blocked, these seemingly permanently depressed worms cease to be depressed (Dagenhardt et al., 2017). They return to moving around and responding to food cues. Fascinatingly, this serotonin depression cure seems to work all the way from C. elegans to humans. The primary medications to treat depression in humans are selective serotonin reuptake inhibitors (SSRIs), such as Prozac, which evidence suggests reduce the level of serotonin in the brain. There is complexity here. At first, SSRIs actually increase the level of serotonin in the brain by blocking the reuptake of serotonin from synapses. But over the course of weeks, this blocking changes the responses of serotonin neurons and makes them reduce their signaling, hence the net effect is to reduce serotonin levels. This is why SSRIs can make depression worse at first, but over the course of two to six weeks, people start to feel better. Of course, there is still controversy around whether this story is exactly right.
给抑郁小鼠服用阿片类药物可以减轻其抑郁迹象（Berrocoso 等人，2013 年；Zomkowski 等人，2005 年）。
Giving opioids to depressed mice reduces their signs of depression (Berrocoso et al., 2013; Zomkowski et al., 2005).
不可避免的压力源的存在：作为进一步的证据，请考虑：是什么加速了蠕虫放弃的速度？能量储备。如果能量储备较低（如胰岛素水平低所表明的那样），蠕虫会更快地放弃（Skora 等人，2018 年）。这似乎过于简单，无法提供有关人类抑郁或压力的信息，但您可能会对其中的联系感到惊讶。胰岛素信号与人类抑郁和慢性压力有着众所周知的联系，尽管仍然神秘：糖尿病患者（一种胰岛素信号被破坏的疾病）的抑郁率是一般人群的三倍（Anderson 等人，2001 年；Gavard 等人，1993 年）；即使没有糖尿病，抑郁症患者也表现出比普通人群更高的胰岛素抵抗普通人群（Hung 等人，2007；Kan 等人，2013），即使没有典型抑郁症，糖尿病患者仍报告冷漠/快感缺乏是一种常见症状（Bruce 等人，2015；Carter 和 Swardfager，2016）。此外，糖尿病小鼠比非糖尿病小鼠表现出更多的抑郁症状，这种影响可以通过注射胰岛素完全逆转（Gupta 等人，2014）。如果你对秀丽隐杆线虫进行基因编辑以减少其胰岛素信号，它会变得永久性抑郁，甚至在响应食物信号时也会表现出运动量急剧减少（Dagenhardt 等人，2017）。快感缺乏是慢性压力的一个进化有益特征——它在困难面前放弃以节省能量。胰岛素似乎是动物有多少能量储备的关键信号，因此动物在困难面前变得长期紧张的可能性有多大。
presence of inescapable stressors: As further evidence for this, consider: What accelerates the speed at which a worm gives up? Energy reserves. If energy reserves are low, as signaled by low levels of insulin, worms give up much faster (Skora et al., 2018). This may seem too simplistic to be informative about depression or stress in humans, but you might be surprised by the connections. Insulin signaling has a well-known, albeit still mysterious, connection to depression and chronic stress in humans: People with diabetes (a disease where insulin signaling is disrupted) have a three times higher rate of depression than the general population (Anderson et al., 2001; Gavard et al., 1993); people with depression even without diabetes still show higher insulin resistance than the general population (Hung et al., 2007; Kan et al., 2013), and people with diabetes even without classical depression still report apathy/anhedonia as a common symptom (Bruce et al., 2015; Carter and Swardfager, 2016). Further, diabetic mice display more signs of depression than nondiabetic mice, an effect that is completely reversed by insulin administration (Gupta et al., 2014). If you genetically edit C. elegans to reduce its insulin signaling, it becomes permanently depressed, showing dramatically reduced movement even in response to food signals (Dagenhardt et al., 2017). Anhedonia is an evolutionary beneficial feature of chronic stress—it is giving up to save energy in the face of hardship. And insulin seems to be a key signal for how much energy reserve an animal has, hence how likely an animal is to become chronically stressed in the face of hardship.
刺激已经消失： AJ Hill 等人，2014 年。
stimuli have gone away: A. J. Hill et al., 2014.
蛞蝓和果蝇：有关蟑螂的习得性无助，请参阅 GE Brown, Anderson, et al., 1994；GE Brown and Stroup, 1988。有关蛞蝓，请参阅 GE Brown, Davenport, et al., 1994。有关果蝇，请参阅 GE Brown et al., 1996。
slugs, and fruit flies: For learned helplessness in cockroaches, see G. E. Brown, Anderson, et al., 1994; G. E. Brown and Stroup, 1988. For slugs, see G. E. Brown, Davenport, et al., 1994. For fruit flies, see G. E. Brown et al., 1996.
甚至根本没有血清素神经元：刺胞动物体内的运动回路很可能也受到各种神经肽的调节，但在刺胞动物中，这种决定似乎不是由与两侧对称动物相同的神经调节剂驱动的。在刺胞动物中，多巴胺已被证明可以抑制进食反应（Hanai 和 Kitajima，1984 年）和触手收缩（Hudman 和 McFarlane，1995 年），甚至引发睡眠（Kanaya 等人，2020 年）。许多刺胞动物中不存在血清素（Carlberg 和 Anctil，1993 年；Takeda 和 Svendsen，1991 年），在发现血清素的少数物种中，它似乎主要诱导产卵（Tremblay 等人，2004 年）。去甲肾上腺素会增加刺丝囊释放的可能性（Kass-Simon 和 Pierobon，2007 年），并改变水母蠕动波的速度（Pani 等人，1995 年）。在大多数情况下，这些神经调节剂似乎参与调节反射，而不是像两侧对称动物那样直接触发导航行为。
don’t even have serotonin neurons at all: It is likely that the locomotive circuits within Cnidarians are also modulated by various neuropeptides, but in Cnidarians, such decisions don’t seem to be driven by the same neuromodulators as in Bilateria. In Cnidaria, dopamine has been shown to inhibit the feeding response (Hanai and Kitajima, 1984) and tentacle contraction (Hudman and McFarlane, 1995) and even trigger sleep (Kanaya et al., 2020). Serotonin is not present across many Cnidarians (Carlberg and Anctil, 1993; Takeda and Svendsen, 1991), and in the few species where it is found, it seems to primarily induce spawning (Tremblay et al., 2004). Norepinephrine increases the likelihood of nematocyst release (Kass-Simon and Pierobon, 2007) and changes the speed of peristalsis waves in medusa (Pani et al., 1995). For the most part, these neuromodulators seem involved in modulating reflexes, and not, as in Bilateria, directly triggering behavioral repertoires for navigation.
第四章：联想、预测和学习的曙光
Chapter 4: Associating, Predicting, and the Dawn of Learning
“我们什么都不是”： Seeger，2009年。
“it we are nothing”: Seeger, 2009.
“错误来源”： Todes，2014。
“source of error”: Todes, 2014.
“引发狗对食物的想法”：同上。
“elicit in the dog thoughts about food”: Ibid.
仅对声音作出反应而踢腿： Irwin，1943 年；Twitmyer，1905 年。
kick in response to just the sound: Irwin, 1943; Twitmyer, 1905.
脊髓中的简单回路： Illich 等人，1994 年。
simple circuits in their spinal cords: Illich et al., 1994.
远离盐： Amano 和 Maruyama，2011；佐伯等人，2001；富冈等人，2006；温等人，1997。
away from the salt: Amano and Maruyama, 2011; Saeki et al., 2001; Tomioka et al., 2006; Wen et al., 1997.
食物气味： Morrison 等人，1999 年；Y. Zhang 等人，2005 年。
from that food smell: Morrison et al., 1999; Y. Zhang et al., 2005.
朝向该温度： Hedgecock 和 Russell，1975 年。
toward that temperature: Hedgecock and Russell, 1975.
会持续数天：这种效应不仅仅是对戒断反应的敏感化；如果你电击一只蛞蝓，但没有将电击与水龙头配对，尽管之前电击次数相同，但它不会像以前那样大量地抽离水龙头。（Carew 等人，1981a；1981b；Walters 等人，1981）。有关简单学习电路的综述，请参阅 Hawkins & Kandel 1984。）
will last for days: This effect is not just a sensitization of the withdrawal response; if you shock a slug but do not pair the shock with the tap, it does not withdraw as much to the tap despite the same number of prior shocks. (Carew et al., 1981a; 1981b; Walters et. al. 1981). For a review of simple learning circuits see Hawkins & Kandel 1984.)
能够学习联想：我只知道一份关于刺胞动物的联想学习的报告，其中海葵学会了收缩触手以响应预示着电击的光，这是 Haralson 等人在 1975 年进行的一项实验。其他尝试没有复制这一结果（Rushforth，1973 年）。Torley，2009 年，进行了文献检索并与刺胞动物行为专家进行了多次个人调查，但没有找到证实刺胞动物中经典条件作用的研究。Ginsburg 和 Jablonka，2019 年，同样得出结论，刺胞动物不表现出联想学习。
capable of learning associations: I am aware of only one report of associative learning in Cnidaria, where a sea anemone learned to contract its tentacles in response to a light that predicted shock, an experiment by Haralson et al., 1975. Other attempts have not replicated this result (Rushforth, 1973). Torley, 2009, performed a literature search and engaged in numerous personal inquiries with experts in Cnidarian behavior and did not find studies confirming classical conditioning in Cnidaria. Ginsburg and Jablonka, 2019, similarly conclude that Cnidaria do not exhibit associative learning.
一个称为灭绝的过程：巴甫洛夫，1927年。
a process called extinction: Pavlov, 1927.
古埃迪卡拉海： Ruben 和 Lukowiak 1982。
ancient Ediacaran Sea: Ruben and Lukowiak 1982.
信用分配问题受到阻塞： Kamin，1969。
credit assignment problem was blocking: Kamin, 1969.
在动物界：蜜蜂（Abramson 和 Bitterman，1986 年）、软体动物（Loy 等人，2006 年）、鱼类（Mitchell 等人，2011 年）、山羊（Lubow 和 Moore，1959 年）和大鼠（Ackil 等人，1969 年；Boughner 和 Papini，2006 年）中都存在潜在抑制。在扁虫 (Prados、Alvarez、Howarth 等，2013)、蜜蜂 (Couvillon 和 Bitterman，1989)、软体动物 (Acebes 等，2009；Sahley 等，1981)、大鼠 (Prados、Alvarez、Acebes 等，2013)、人类 (Prados、Alvarez、Acebes 等，2013)、兔子 (Merchant 和 Moore，1973) 和猴子 (Cook 和 Mineka，1987) 中都观察到了遮蔽和阻塞现象。
in the animal kingdom: Latent inhibition has been shown in honeybees (Abramson and Bitterman, 1986), mollusks (Loy et al., 2006), fish (Mitchell et al., 2011), goats (Lubow and Moore, 1959), and rats (Ackil et al., 1969; Boughner and Papini, 2006). Overshadowing and blocking have been observed in flatworms (Prados, Alvarez, Howarth, et al., 2013), honeybees (Couvillon and Bitterman, 1989), mollusks (Acebes et al., 2009; Sahley et al., 1981), rats (Prados, Alvarez, Acebes, et al., 2013), humans (Prados, Alvarez, Acebes, et al., 2013), rabbits (Merchant and Moore, 1973), and monkeys (Cook and Mineka, 1987).
阻挡和遮蔽： Illich 等人，1994 年。
blocking, and overshadowing: Illich et al., 1994.
记忆是这些印象：伯纳姆，1888年。
memories were these impressions: Burnham, 1888.
“曾经被折叠”： Levy，2011。
“has once been folded”: Levy, 2011.
是持续的“振动”： Burnham，1889年。
were persistent “vibrations”: Burnham, 1889.
学习如何进行：有关突触学习的不同发现的历史的良好回顾，特别是基于时间的学习规则，请参阅 Markram 等人，2011 年的文章。
how learning might work: For a good review of the history of different discoveries on synaptic learning, in particular timing-based learning rules, see Markram et al., 2011.
参与赫布学习： Ramos-Vicente, D. et al., 2018；斯特罗贝尔，D.2021。
involved in Hebbian learning: Ramos-Vicente, D. et al., 2018; Stroebel, D. 2021.
第五章：寒武纪生命大爆发
Chapter 5: The Cambrian Explosion
极其相似： Grillner 和 Robertson，2016 年。
are incredibly similar: Grillner and Robertson, 2016.
许多动物物种： P.Chance，1999。
many animal species: P. Chance, 1999.
图 5.5：Thorndike 的拼图盒之一：图片来自 Thorndike，1898 年（图 1）。
Figure 5.5: One of Thorndike’s puzzle boxes: Picture is from Thorndike, 1898 (figure 1).
“再次陷入那种境地”： P. Gray，2011年。
“again in that situation”: P. Gray, 2011.
获取食物的特定按钮： Adron 等，1973 年。
specific button to get food: Adron et al., 1973.
陷入网中： C. Brown，2001 年；C. Brown 和 Warburton，1999 年。
getting caught in a net: C. Brown, 2001; C. Brown and Warburton, 1999.
第六章：时间差分学习的演变
Chapter 6: The Evolution of Temporal Difference Learning
不去工作：明斯基，1961 年。
not going to work: Minsky, 1961.
使用预测奖励的行为：公平地说，这些想法中的一些已经存在于运筹学领域的动态规划中。Sutton 的贡献是实现了可以同时解决策略和价值函数。
behaviors using predicted rewards: To be fair, some of these ideas were already present in dynamic programming in the operations research world. Sutton’s contribution was realization you could solve the policy and the value function simultaneously.
使用时间差异学习的西洋双陆棋：请注意，TD-Gammon 不是演员-评论家模型，而是时间差异学习的简单版本，它直接学习所谓的价值函数。但时间差异的引导原理是相同的。
backgammon using temporal difference learning: Note that TD-Gammon was not an actor-critic model but a simpler version of temporal difference learning that learned what is called the value function directly. But the principle of bootstrapping on temporal differences was the same.
“惊人的性能水平”： Tesauro，1994。
“staggering level of performance”: Tesauro, 1994.
连续二十四小时： Olds，1956 年；Olds 和 Milner，1954 年。这些实验实际上刺激了隔膜区域，从而触发了多巴胺的释放。后来的实验证实，事实上，是多巴胺介导了隔膜刺激的效果；如果你给老鼠注射多巴胺阻断药物，老鼠将不再推动隔膜刺激的杠杆（Wise，2008 年评论）。
for twenty-four hours straight: Olds, 1956; Olds and Milner, 1954. These experiments actually stimulated the septal area, which triggered dopamine release. Later experiments confirmed that it was, in fact, dopamine that mediated the effect of septal stimulation; if you inject a rat with dopamine-blocking drugs, the rat will no longer push a lever for septal stimulation (reviewed in Wise, 2008).
有利于多巴胺刺激： “一只饥饿的老鼠常常忽略现有的食物，而追求通过电刺激自己来获得快感”（Olds，1956）。
favor of dopamine stimulation: “A hungry rat often ignored available food in favor of the pleasure of stimulating itself electrically” (Olds, 1956).
（反复从水中取出）： Kily 等人，2008 年。
(repeatedly removed from water): Kily et al., 2008.
此类增强多巴胺的化学物质： Cachat 等人，2010 年；Gerlai 等人，2000 年、2009 年。
such dopamine-enhancing chemicals: Cachat et al., 2010; Gerlai et al., 2000, 2009.
时间差异学习信号： Schultz 等人，1997 年。
temporal difference learning signal: Schultz et al., 1997.
十六秒内吃完食物：小林和舒尔茨，2008 年。
food in sixteen seconds: Kobayashi and Schultz, 2008.
“预测和奖励的神经基础”： Schultz 等人，1997 年。
“A Neural Substrate of Prediction and Reward”: Schultz et al., 1997.
猴子和人类的大脑： Grillner 和 Robertson，2016 年；JM Li，2012 年；Vindas 等人，2014 年。
monkey, and human brains: Grillner and Robertson, 2016; J. M. Li, 2012; Vindas et al., 2014.
对于未来奖励的预测失败：有关此想法的更多信息，请参阅 Joseph LeDoux 在 LeDoux 等人于 2017 年撰写的关于回避学习的精彩评论。
for a failed prediction of future reward: For more information on this idea, see Joseph LeDoux’s great review of avoidance learning in LeDoux et al., 2017.
预期的奖励或惩罚：很难区分巴甫洛夫式学习和从忽略中学习，而且关于回避学习的机制一直存在争议。有关一项表明鱼类确实从忽略冲击中学习的出色研究，请参阅 Zerbolio 和 Royalty，1982 年。
expected rewards or punishments: It can be hard to differentiate Pavlovian learning from learning from omission, and there are ongoing debates about the mechanisms of avoidance learning. For a great study showing that fish do truly learn from the omission of shock, see Zerbolio and Royalty, 1982.
光照后五秒： MR Drew 等人，2005 年；A. Lee 等人，2010 年。Cheng 等人，2011 年评论。
five seconds after light: M. R. Drew et al., 2005; A. Lee et al., 2010. Reviewed in Cheng et al., 2011.
一天的周期： Eelderink-Chen 等人，2021 年，展示了原核生物的昼夜节律；McClung，2006 年，展示了植物的昼夜节律。
cycle of the day: Eelderink-Chen et al., 2021, shows circadian rhythm in prokaryotes; McClung, 2006, shows circadian rhythm in plants.
事件之间的间隔： Abramson 和 Feinman，1990 年；Craig 等人，2014 年。Abramson 和 Wells，2018 年评论。
intervals between events: Abramson and Feinman, 1990; Craig et al., 2014. Reviewed in Abramson and Wells, 2018.
最大化多巴胺释放：皮层和基底神经节之间的连接包含不同类型的多巴胺受体。多巴胺的快速激增会增强皮层神经元和特定基底神经节神经元之间的联系，从而驱动执行（或取消）特定动作，并削弱皮层神经元和另一组基底神经节神经元之间的联系，从而驱动停止（或关闭）特定动作。多巴胺的快速下降会产生相反的效果。通过这些平行回路，多巴胺爆发会强化最近的行为，使它们更有可能再次发生，而多巴胺下降会惩罚行为，使它们不太可能再次发生。
that maximize dopamine release: The connections between the cortex and basal ganglia contain different types of dopamine receptors. Rapid spikes of dopamine drive a strengthening between the cortical neurons and specific basal ganglia neurons that drive doing (or ungating) specific actions, and a weakening between cortical neurons and a different set of basal ganglia neurons that drive stopping (or gating) specific actions. Rapid declines of dopamine have the opposite effect. Through these parallel circuits, dopamine bursts reinforce recent behaviors and make them more likely to reoccur and dopamine declines punish behaviors and make them less likely to reoccur.
多巴胺基底神经节： Cone 等人，2016 年，证明钠食欲的价态变化源于从下丘脑外侧到中脑多巴胺神经元的输入变化。
basal ganglia of dopamine: Cone et al., 2016, shows evidence that changing valence of sodium appetite derives from changes in input from lateral hypothalamus to midbrain dopamine neurons.
直接多巴胺神经元：所谓的演员回路从纹状体（基底神经节的输入结构）的基质流出；所谓的批评回路从纹状体的纹状体流出。
dopamine neurons directly: The purported actor circuit flows from the matrix of the striatum (the input structure of the basal ganglia); the purported critic circuit flows from the striosomes of the striatum.
第 7 章：模式识别问题
Chapter 7: The Problems of Pattern Recognition
特定类型的分子： Niimura，2009，提供了进化证据，表明现代脊椎动物的嗅觉受体起源于早期脊椎动物的早期脊索动物。文昌鱼（一种经常用作早期脊索动物模型的动物）有 31 个功能性脊椎动物类嗅觉受体基因，七鳃鳗（经常用作早期脊椎动物的模型）有 32 个脊椎动物类嗅觉受体基因。请注意，不同的谱系扩大了嗅觉受体的数量；一些现代鱼类有超过一百个，而大鼠有超过一千个。
specific types of molecules: Niimura, 2009, provides evolutionary evidence that the olfactory receptors in modern vertebrates originated just before early vertebrates in the first chordates. Amphioxus (an animal often used as a model of the first chordates) has thirty-one functional vertebrate-like olfactory receptor genes and the lamprey (often used as a model for the early vertebrates) has thirty-two vertebrate-like olfactory receptor genes. Note that different lineages expanded the number of olfactory receptors; some modern fish have over one hundred, and rats have over one thousand.
正如 Niimura, 2012 所讨论的，虽然无脊椎动物也有嗅觉受体，但它们的嗅觉受体似乎是独立进化的：“[嗅觉受体] 基因也从其他无脊椎动物中鉴定出来，包括昆虫、线虫、棘皮动物和软体动物……然而，它们的进化起源与脊椎动物 [嗅觉受体] 基因不同。昆虫和脊椎动物嗅觉系统的神经解剖学特征很常见，但昆虫和脊椎动物 [嗅觉受体] 基因彼此截然不同，并且没有序列相似性。”
As discussed in Niimura, 2012, while invertebrates also have olfactory receptors, their olfactory receptors seem to have independently evolved: “[Olfactory receptor] genes were also identified from other invertebrates including insects, nematodes, echinoderms, and mollusks . . . however, their evolutionary origins are distinct from that of vertebrate [olfactory receptor] genes. The neuroanatomical features of insect and vertebrate olfactory systems are common, but insect and vertebrate [olfactory receptor] genes are strikingly different to each other and share no sequence similarities.”
相似但不相同： DA Wilson（2009 年）和 Laurent（1999 年）提出了类似的嗅觉编码模型。Barnes 等人（2008 年）在大鼠嗅觉皮层中发现了模式完成的证据。Yaksi 等人（2009 年）在鱼类中发现了类似类型的模式分离和完成的证据。有关提出三层皮层执行这种类型的自我关联的原始论文之一，请参阅 Marr（1971 年）。
are similar but not the same: D. A. Wilson, 2009, and Laurent, 1999, proposed a similar model of olfactory coding. Barnes et al., 2008, found evidence of pattern completion in olfactory cortex of rats. Yaksi et al., 2009, found evidence of similar types of pattern separation and completion in fish. For one of the original papers suggesting that the three-layered cortex performs this type of auto-association, see Marr, 1971.
人工神经网络：图片来自https://en.wikipedia.org/wiki/Artificial_neural_network#/media/File:Colored_neural_network.svg。
An artificial neural network: Picture from https://en.wikipedia.org/wiki/Artificial_neural_network#/media/File:Colored_neural_network.svg.
三层神经元：硬骨鱼可能没有清晰的层次，但七鳃鳗的皮层是分层的，爬行动物也是如此，因此我继续假设早期脊椎动物的皮层是分层的（Suryanarayana 等人，2022 年）。
three layers of neurons: Teleost fish may not have clear layers, but the lamprey has a layered cortex, as do reptiles, hence I proceed with the assumption that the cortex of early vertebrates was layered (Suryanarayana et al., 2022).
网络进行数学运算： McCloskey 和 Cohen，1989 年。有关持续学习当前挑战的评论，请参阅 Parisi 等人，2019 年，以及 Chen 和 Liu，2018 年。
networks to do math: McCloskey and Cohen, 1989. For reviews of current challenges in continual learning see Parisi et al., 2019, and Chen and Liu, 2018.
前一年：布朗，2001 年。
the year before: Brown, 2001.
学习这种新模式： Grossberg，2012。
learn this new pattern: Grossberg, 2012.
向猫展示不同的视觉刺激： Hubel 和 Wiesel，1959、1962、1968。
presented the cats with different visual stimuli: Hubel and Wiesel, 1959, 1962, 1968.
图 7.9：图来自 Manassi 等人，2013 年。已获许可使用。
Figure 7.9: Figure from Manassi et al., 2013. Used with permission.
由 Hubel 和 Wiesel 发现：福岛，1980 年。
discovered by Hubel and Wiesel: Fukushima, 1980.
点击图片获取食物： Wegman 等人，2022 年。
tap pictures to get food: Wegman et al., 2022.
旋转和平移的物体： Worden 等人，2021 年。皮层和丘脑紧密相连。最初人们认为丘脑仅仅是向皮层输入的“中继”。但新的研究开始对此提出质疑。三个观察结果提示我们，丘脑和皮层之间的相互作用可能对不变性问题很重要。首先，大多数进入皮层的感觉输入都流经丘脑——来自眼睛、耳朵和皮肤的输入都首先进入丘脑，然后分支到皮层的各个区域。然而，这条规则有一个例外：嗅觉。唯一跳过丘脑并直接连接到皮层的感觉是嗅觉。也许这是因为嗅觉是唯一没有不变性问题的感觉；它不需要与丘脑相互作用来识别不同尺度、旋转和平移的物体。其次，丘脑和皮层是一起进化的。即使是在最遥远的脊椎动物中，比如七鳃鳗，也存在皮层和丘脑，并且它们之间的相互作用与其他脊椎动物相似。这表明它们的功能可能是从它们之间的相互作用中产生的。第三，丘脑的回路似乎精确地控制和引导皮层不同区域之间的连接。
rotated and translated objects: Worden et al., 2021. The cortex and thalamus are densely interconnected. It was originally believed that the thalamus was merely a “relay” of input to the cortex. But new research is beginning to call this into question. Three observations give us hints that the interactions between the thalamus and the cortex might be important for the invariance problem. First, most sensory input to the cortex flows through the thalamus—input from eyes, ears, and skin all first go to the thalamus and then branch out to various regions of the cortex. However, there is one exception to this rule: smell. The single sense that skips the thalamus and connects directly to the cortex is smell. Perhaps this is because smell is the only sense that does not have the invariance problem; it does not need to interact with the thalamus to recognize objects at different scales, rotations, and translations. Second, the thalamus and the cortex evolved together. Even in the most distant vertebrates, like the lamprey fish, there is both a cortex and a thalamus with similar interactions between them as in other vertebrates. This suggests that their function might be emergent from interactions between them. And third, the circuitry of the thalamus seems to precisely gate and route connections between different areas of the cortex.
第 8 章：生命为何变得奇妙
Chapter 8: Why Life Got Curious
蒙特祖玛的复仇：这个故事于 2020 年基督教中重述。
Montezuma’s Revenge: This story is retold in Christian, 2020.
对人类婴儿的好奇心：有关对鱼类的好奇心，请参阅 Budaev, 1997。有关对小鼠的好奇心，请参阅 Berlyne, 1955。有关对猴子的好奇心，请参阅 Butler and Harlow, 1954。有关对人类婴儿的好奇心，请参阅 Friedman, 1972。
to human infants: For curiosity in fish, see Budaev, 1997. For curiosity in mice, see Berlyne, 1955. For curiosity in monkeys, see Butler and Harlow, 1954. For curiosity in human infants, see Friedman, 1972.
没有“真正的”奖励：松本和彦坂，2009 年。
no “real” reward: Matsumoto and Hikosaka, 2009.
头足类动物表现出好奇心：评论于 Pisula，2009 年。有关蟑螂的好奇心，请参阅 Durier 和 Rivault，2002 年。有关蚂蚁的好奇心，请参阅 Godzinska，2004 年。有关头足类动物的好奇心，请参阅 Kuba 等人，2006 年。
cephalopods, show curiosity: Reviewed in Pisula, 2009. For cockroach curiosity, see Durier and Rivault, 2002. For curiosity in ants, see Godzinska, 2004. For cephalopod curiosity, see Kuba et al., 2006.
在早期的两侧对称动物中并不存在：在 2009 年 Pisula 对好奇心的评论中，第 48 页，他得出结论：“探索行为的相似性必定是趋同的结果，即对环境提出的类似挑战做出类似的反应，而不是共同的祖先。”
wasn’t present in early bilaterians: In the review on curiosity by Pisula, 2009, on page 48, he concludes that “parallels in exploratory behaviors must therefore be a result of convergence, i.e. a similar response to similar challenges presented by the environment, rather than common ancestry.”
鱼类也表现出这种效应： 1964 年，Wertheim 和 Singer 在金鱼身上发现了这种部分强化效应。1962 年，Gonzales 等人也证明了这一点，尽管与哺乳动物身上的这种效应有一些不同。
Fish exhibit this effect as well: This partial reinforcement effect was found in goldfish in Wertheim and Singer, 1964. It was also shown in Gonzales et al., 1962, although with some differences from how the effect works in mammals.
第九章 世界的第一个模型
Chapter 9: The First Model of the World
直接放入装有食物的容器中： Durán 等，2008 年，2010 年。
directly to the container with the food: Durán et al., 2008, 2010.
一件事物相对于另一件事物： Larsch 等人，2015 年。有关涡虫，请参阅 Pearl，1903 年。有关蜜蜂，请参阅 Abramson 等人，2016 年。
one thing relative to another thing: Larsch et al., 2015. For planarians see Pearl, 1903. For bees see Abramson et al., 2016.
再次穿过整个循环： Wehner 等人，2006 年。
through the entire loop again: Wehner et al., 2006.
图 9.1：图片由 Carlyn Iverson / Science Source 提供。已获许可使用。
Figure 9.1: Image by Carlyn Iverson / Science Source. Used with permission.
面向某个方向： Bassett 和 Taube，2001。
facing a certain direction: Bassett and Taube, 2001.
腹侧皮质和内侧皮质：注意如何细分存在争议鱼类和两栖动物的皮质部分。有人认为有四个区域，另外一个是背部皮质（见 Striedter 和 Northcutt，2019 年）。
ventral cortex, and medial cortex: Note there is controversy around how to subdivide parts of the cortex in fish and amphibians. Some argue there are four areas, an additional one being the dorsal cortex (see Striedter and Northcutt, 2019).
早期脊椎动物的皮质：请注意，这些皮质区域的位置在某些现代脊椎动物中发生了变化，这导致对同一功能结构的不同命名，为了便于阅读，我故意省略了这种复杂性。例如，在硬骨鱼中，皮质的折叠方式与七鳃鳗、爬行动物和哺乳动物不同（在硬骨鱼中，皮质经历外翻，向外折叠，而不是内翻，向内折叠）。因此，相同功能结构（皮质的海马部分）的位置最终位于大脑的不同位置，因此解剖学家对这些区域的名称也不同。当我提到早期脊椎动物的内侧皮质时，我指的是早期脊椎动物中的功能结构，后来成为哺乳动物的海马体。在七鳃鳗和爬行动物中，皮质的这一部分是内侧皮质，而在硬骨鱼中，皮质的相同部分是外侧皮质。为简单起见，我将皮质的这一区域称为海马体，并以图形方式仅表示内陷状态下的皮质（因此海马体结构出现在内侧部分），因为我们使用七鳃鳗作为早期脊椎动物的模型生物。
cortex of early vertebrates: Note that the location of these areas of cortex is shifted in some modern vertebrates, which leads to different naming of the same functional structure, a complexity I have intentionally omitted for readability. In teleost fish, for example, the cortex does not fold in the same ways as in the lamprey, reptiles, and mammals (in teleost fish, the cortex goes through evagination, folding outward, instead of invagination, folding inward). So the location of the same functional structure (the hippocampal part of the cortex) ends up in different places in the brain, and thus the names anatomists give these areas differ. When I refer to the medial cortex of early vertebrates, I am referring to the functional structure in early vertebrates that would later become the hippocampus in mammals. In the lamprey and reptiles, this part of the cortex is the medial cortex, whereas in teleost fish, the identical part of the cortex is the lateral cortex. For simplicity I refer to this area of cortex as hippocampus and graphically represent the cortex only in its invaginated state (hence the hippocampus structure shows up in the medial part), since we use the lamprey as the model organism for early vertebrates.
面临着具体的方向： Fotowat et al., 2019；维内平斯基等人，2020。
are facing specific directions: Fotowat et al., 2019; Vinepinsky et al., 2020.
空间地图： Petrucco 等人，2022 年，展示了鱼类后脑中存在头部方向细胞的证据。有关啮齿动物头部方向细胞的整体网络和海马输入的综述，请参阅 Yoder 和 Taube，2014 年。
a spatial map: Petrucco et al., 2022, shows evidence of head-direction cells in the hindbrain of fish. For a review of the overall network of head-direction cells and input to the hippocampus in rodents, see Yoder and Taube, 2014.
记住地点的能力： Broglio et al., 2010；杜兰等人，2010；洛佩斯等人，2000；罗德里格斯等人，2002。
ability to remember locations: Broglio et al., 2010; Durán et al., 2010; López et al., 2000; Rodríguez et al., 2002.
迷宫中的转弯： Rodríguez 等人，2002 年。
turn in a maze: Rodríguez et al., 2002.
获取食物的空间： Durán 等，2010。
space to get food: Durán et al., 2010.
给定不同的起始位置： Broglio 等人，2010 年。
given different starting locations: Broglio et al., 2010.
同样会损害空间导航： Naumann 和 Laurent，2016 年；Peterson，1980 年；Rodríguez 等人，2002 年。
similarly impairs spatial navigation: Naumann and Laurent, 2016; Peterson, 1980; Rodríguez et al., 2002.
第 10 章：神经黑暗时代
Chapter 10: The Neural Dark Ages
到以前不适合居住的地区： Algeo，1998。
to previously inhospitable areas: Algeo, 1998.
were thirty meters tall: Beck, 1962.
晚泥盆世灭绝： Algeo 等人，1995 年；McGhee，1996 年。
The Late Devonian Extinction: Algeo et al., 1995; McGhee, 1996.
比在水下能做到的还要多： Mugan 和 MacIver，2020 年。
than you can underwater: Mugan and MacIver, 2020.
和计划是鸟类： Boeckle 和 Clayton，2017 年。
and plan are birds: Boeckle and Clayton, 2017.
图 10.3：我省略了早期脊椎动物的背部皮质，因为关于其存在仍存在争议（Striedter 和 Northcutt，2019 年）。有关内侧皮质与海马体、外侧皮质与嗅觉皮质以及腹侧皮质与联想杏仁核之间的对齐关系，请参阅 Luzzati，2015 年；Striedter 和 Northcutt，2019 年。
Figure 10.3: I am leaving out the dorsal cortex from early vertebrates because there is still debate about its presence (Striedter and Northcutt, 2019). For the alignment between medial cortex to hippocampus, lateral cortex to olfactory cortex, and ventral cortex to associative amygdala, see Luzzati, 2015; Striedter and Northcutt, 2019.
第 11 章：生成模型和新皮层之谜
Chapter 11: Generative Models and the Neocortical Mystery
刺激引起什么反应： Talbot 等人，1968 年。
stimuli elicited what responses: Talbot et al., 1968.
Mountcastle 假设的检验： von Melchner 等人，2000 年。
test of Mountcastle’s hypothesis: von Melchner et al., 2000.
图 11.4： Jastrow (1899) 的“编辑”图像。其他图像来自 Lehar S. (2003)，取自维基百科。
Figure 11.4: “Editor” image from Jastrow (1899). Other images from Lehar S. (2003), obtained from Wikipedia.
图 11.5： Schroeder (1858) 的楼梯。Louis Necker (1832) 的“内克尔立方体”。Jastrow (1899) 的鸭子或兔子。
Figure 11.5: Staircase from Schroeder (1858). “Necker cube” from Louis Necker (1832). Duck or rabbit from Jastrow (1899).
图 11.6：图像来自 Fahle 等人，2002 年。经麻省理工学院出版社许可使用。
Figure 11.6: Image from Fahle, et al., 2002. Used with permission by The MIT Press.
与亥姆霍兹理论：亥姆霍兹思想的后期体现包括综合分析（Neisser, 1967）和 Mackay 的认识论自动机（MacKay, 1956）。
with Helmholtz’s theory: Later incarnations of Helmholtz’s ideas include analysis-by-synthesis (Neisser, 1967) and Mackay’s epistemological automata (MacKay, 1956).
亥姆霍兹机： Dayan，1997；Hinton 等人，1995。
the Helmholtz machine: Dayan, 1997; Hinton et al., 1995.
自学成才： Dayan，1997；同上。
learned on its own: Dayan, 1997; Ibid.
图 11.8： Hinton 等人于 1995 年拍摄的图像。已获许可使用。
Figure 11.8: Image from Hinton et al., 1995. Used with permission.
thispersondoesnotexist.com：使用 StyleGAN2（Karras 等人，2020）。
thispersondoesnotexist.com: Uses a StyleGAN2 (Karras et al., 2020).
图 11.10：图来自 He 等人，2019 年。已获许可使用。
Figure 11.10: Figure from He et al., 2019. Used with permission.
使用生成模型： Reichert 等人，2013 年。
with a generative model: Reichert et al., 2013.
“受限幻觉”： Seth，2017 年。
“constrained hallucination”: Seth, 2017.
存在 REM 睡眠：唯一被证明也具有类似哺乳动物的睡眠状态（即 NREM 和 REM 睡眠交替）（暗示做梦）的非哺乳动物是鸟类（Johnsson 等人，2022 年；Lesku 和 Rattenborg，2014 年；Rattenborg 等人，2022 年）。
presence of REM sleep: The only nonmammalian animals shown to also have mammal-like sleep states of alternated NREM and REM sleep (suggestive of dreaming) are birds (Johnsson et al., 2022; Lesku and Rattenborg, 2014; Rattenborg et al., 2022).
强制生成的过程：有关我们做梦的原因、与生成模型的联系以及做梦的其他解释的理论的详细概述，请参阅 Prince and Richards，2021 年。
process of forced generation: For a nice overview of theories of why we dream, the connection to generative models, and alternative explanations for dreaming, see Prince and Richards, 2021.
处理实际视觉数据： van der Wel 和 van Steenbergen，2018。
processing actual visual data: van der Wel and van Steenbergen, 2018.
想象同样的事情： O'Craven 和 Kanwisher，2000 年。
imagine the same thing: O’Craven and Kanwisher, 2000.
移动身体部位： Doll 等，2015 年。
moving the body parts: Doll et al., 2015.
（记录他们的大脑）：有关详细综述，请参阅 Pearsona 和 Kosslynb，2015 年。现在有强有力的证据表明，当人们进行视觉化时，V1 区会出现活动（Albers 等人，2013 年；Slotnick 等人，2005 年；Stokes 等人，2009 年）。有关人们可以从视觉新皮层中的激活中解码图像的证据，请参阅 Kay 等人，2008 年；Naselaris 等人，2015 年；Thirion 等人，2006 年。
(recording their brains): For a good review, see Pearsona and Kosslynb, 2015. There is now strong evidence that when one visualizes, there is activity in area V1 (Albers et al., 2013; Slotnick et al., 2005; Stokes et al., 2009). For evidence that one can decode imagery from activation in visual neocortex, see Kay et al., 2008; Naselaris et al., 2015; Thirion et al., 2006.
（左视野）： Bisiach 和 Luzzatti，1978 年；Farah 等人，1992 年。
(the left visual field): Bisiach and Luzzatti, 1978; Farah et al., 1992.
随着时间的推移而展开：杰夫·霍金斯对此有一些精彩的著作。请参阅他的著作《J. Hawkins, 2021》和《J. Hawkins, 2004》。
that unfolds over time: Jeff Hawkins has some great writing about this. See his books J. Hawkins, 2021, and J. Hawkins, 2004.
“来自当今的人工智能系统”： Yann LeCun（@lecun）于 2021 年 12 月 19 日发布推文。
“from AI systems today”: tweeted by Yann LeCun (@lecun) on December 19, 2021.
第 12 章：幻想秀中的老鼠
Chapter 12: Mice in the Imaginarium
旋转、重新缩放或扰动： Volotsky 等人，2022 年。
rotated, rescaled, or perturbed: Volotsky et al., 2022.
在选择方向之前：托尔曼，1939 年，1948 年。
before choosing a direction: Tolman, 1939, 1948.
尝试下一个选项： Steiner 和 Redish，2014。如 Redish，2016 中所述。
try the next option: Steiner and Redish, 2014. As reviewed in Redish, 2016.
考虑替代路径： Schmidt 等人，2013 年。
to consider alternative paths: Schmidt et al., 2013.
右转即可：同上。
turning right and getting: Ibid.
完全没有障碍： Beyiuc，1938（第 409 页）描述了斗鱼的这种发展。Gómez-Laplaza 和 Gerlai，2010，是我所知道的唯一反例，据报道，鱼会对地图进行潜在学习并提前做出正确的决定。如果没有进一步的重复，很难解释单个反例。但如果事实证明某些鱼确实可以进行潜在学习，那么这表明要么早期脊椎动物可以在没有规划的情况下解决潜在学习任务，要么规划在某些鱼中独立进化，要么规划以某种形式存在于早期脊椎动物中。
the barrier at all: Beyiuc, 1938 (page 409), describes this unfolding in a betta fish. Gómez-Laplaza and Gerlai, 2010, is the only counterexample I am aware of where fish were reported to do latent learning of a map and make the right decision ahead of time. It is hard to interpret a single counterexample without further replications. But if it turns out that some fish can, in fact, perform latent learning, then this suggests either that early vertebrates could solve latent-learning tasks without planning, planning evolved independently in some fish, or planning in some form existed in early vertebrates.
绕过障碍： Lucon-Xiccato 等人，2017 年。
navigate around a barrier: Lucon-Xiccato et al., 2017.
跑向盐地： Tindell 等人，2009 年。
runs toward the salt: Tindell et al., 2009.
他们称之为“餐厅街”： Steiner 和 Redish，2014 年。如 Redish，2016 年所评论。另请参阅 Bissonette 等人，2014 年。
they called “restaurant row”: Steiner and Redish, 2014. As reviewed in Redish, 2016. Also see Bissonette et al., 2014.
玩石头、剪刀、布： Abe 和 Lee，2011 年。
play rock, paper, scissors: Abe and Lee, 2011.
也没有发生： D.Lewis，1973年。
did not occur either: D. Lewis, 1973.
（鸟类的推理能力）：有关大鼠因果推理能力的证据，请参阅 Blaisdell 等人，2006 年；MR Waldmann 等人，2012 年。Fischer 和 Schloegl，2017 年得出结论，因果推理能力是在早期哺乳动物中进化而来的，并且在鸟类中也独立进化而来。
(of reasoning in birds): For evidence of causal reasoning in rats, see Blaisdell et al., 2006; M. R. Waldmann et al., 2012. Fischer and Schloegl, 2017, conclude that causal reasoning evolved in early mammals and independently evolved also in birds.
你正在回忆过去： Addis 等人，2007 年。
you’re remembering the past: Addis et al., 2007.
（用想象来看待事物）： O'Craven 和 Kanwisher，2000 年；J. Pearson 等人，2015 年。
(saw with imagining things): O’Craven and Kanwisher, 2000; J. Pearson et al., 2015.
错误的目击者证词： Shermer 等人，2011 年。
mistaken eyewitness testimony: Shermer et al., 2011.
该事件确实发生了： Garry 等人，1996 年。
the event did occur: Garry et al., 1996.
获取更多食物： Crystal，2013；W. Zhou 等，2012。
to get more food: Crystal, 2013; W. Zhou et al., 2012.
用于渲染模拟的结构：许多神经科学家不喜欢使用“情景记忆”这个术语来指代这些较简单的哺乳动物（例如大鼠）回忆过去生活事件的过程。“情景记忆”这个术语包含了许多概念，例如有意识地进行精神时间旅行的经历或某种自传体自我的概念。许多神经科学家改用更安全的术语“情景类记忆”。事实上，尚不清楚大鼠的记忆在经验上有多丰富。但无论如何，来自大鼠的证据表明，情景记忆的前身存在于早期哺乳动物中。
structures for rendering simulations: Many neuroscientists don’t like using the term episodic memory to refer to the process by which these simpler mammals, such as rats, recall past life events. The term episodic memory has been loaded with a baggage of concepts, such as the conscious experience of mentally time traveling or some notion of an autobiographical self. Many neuroscientists instead use the safer term episodic-like memory. Indeed, it isn’t clear how experientially rich the recollections of rats are. But regardless, the evidence from rats suggests that the precursor to episodic memory was present in early mammals.
状态和最佳动作：动作不仅包括运动，还包括“空间中的下一个目标位置”。换句话说，祖先脊椎动物的“空间地图”不被视为基于模型的 RL，因为它似乎不是用于模拟未来动作。但它仍然可以用于位置识别和归巢向量的构建。
state and the best actions: Actions include not just motor movements but the “next target location in space.” In other words, the “spatial map” of ancestral vertebrates is not considered model-based RL because it doesn’t seem to be employed for the purpose of simulating future actions. But it could still be employed for place recognition and the construction of homing vectors.
游戏是无模型的： Mnih 等人，2013 年。
games were model-free: Mnih et al., 2013.
无模型： Wang，2018。
are model-free: Wang, 2018.
第 13 章：基于模型的强化学习
Chapter 13: Model-Based Reinforcement Learning
像国际象棋这样的复杂游戏： Baxter 等人，2000 年，提供了一种 TD 学习方法来下国际象棋（仍然使用树搜索，因此不是无模型的），并很好地总结了在国际象棋中使用无模型方法（无搜索）的困难。
complex games like chess: Baxter et al., 2000, offers a TD-learning approach to playing chess (which still uses tree search and hence is not model-free) and provides a nice summary of the struggles of using model-free approaches (no search) in chess.
围棋冠军李世石：银牌等，2016年。
Go champion Lee Sedol: Silver et al., 2016.
围棋比国际象棋更胜一筹： M.James，2016 年。
Go than in chess: M. James, 2016.
围棋游戏：图片来自https://en.wikipedia.org/wiki/Go_(game)#/media/File:FloorGoban.JPG。
The game of Go: Figure from https://en.wikipedia.org/wiki/Go_(game)#/media/File:FloorGoban.JPG.
围棋的棋盘位置： AlphaGo 纪录片（Kohs，2020 年）。
board position in Go: AlphaGo documentary (Kohs, 2020).
（可能的后续举措）：同上。
(possible subsequent next moves): Ibid.
（正文和导航路径）：同上。
(body and navigational paths): Ibid.
关于任何事物： Devinsky 等人，1995 年；Németh 等人，1988 年；BA Vogt，2009 年。
about anything at all: Devinsky et al., 1995; Németh et al., 1988; B. A. Vogt, 2009.
“没有‘意愿’回答”：达马西奥和范霍森，1983 年。
“no ‘will’ to reply”: Damasio and van Hoesen, 1983.
成为哺乳动物：根据JH Kaas，2011 年的研究，早期哺乳动物的额叶皮质有两个主要区域：前扣带皮层和眶额皮质。当我提到早期哺乳动物的无颗粒前额叶皮质时，我指的是这两个区域。人类的前扣带皮层被认为与啮齿动物的前边缘、下边缘、背侧前扣带皮层同源（Laubach 等人，2018 年；van Heukelum 等人，2020 年），所有这些都可以假定是从第一批哺乳动物的前扣带皮层遗传而来的。
to be a mammal: According to J. H. Kaas, 2011, early mammals had two main areas of frontal cortex: anterior cingulate and orbitofrontal cortex. When I refer to agranular prefrontal cortex of early mammals, I am referring to both of these regions. The anterior cingulate cortex in humans is considered homologous to the prelimbic, infralimbic, dorsal anterior cingulate cortex of rodents (Laubach et al., 2018; van Heukelum et al., 2020), all of which can be assumed to have been inherited from the anterior cingulate cortex of the first mammals.
替代性反复试验：失去头部切换行为：Schmidt 等人，2019 年；失去海马中的目标表征：Ito 等人，2015 年。
vicarious trial and error: Lose head toggling behavior: Schmidt et al., 2019; lose goal representations in hippocampus: Ito et al., 2015.
情景记忆回忆： Frankland 等人，2004 年。
episodic-memory recall: Frankland et al., 2004.
或反事实学习： JL Jones 等人，2012 年。
or counterfactual learning: J. L. Jones et al., 2012.
努力是值得的： Friedman 等人，2015 年。
effort is worth it: Friedman et al., 2015.
重复过去的错误： Frankland 等人，2004 年。
to repeat past mistakes: Frankland et al., 2004.
重复已经完成的动作： Goard 等人，2016 年；Kamigaki 和 Dan，2017 年；Kopec 等人，2015 年。
repeat already completed actions: Goard et al., 2016; Kamigaki and Dan, 2017; Kopec et al., 2015.
耐心获取食物：灭活大鼠额叶皮质的前边缘区域（其 aPFC 的一部分）会增加过早反应（例如，在刺激前释放杠杆）。（Hardung 等人，2017 年；Narayanan 等人，2006 年）。灭活大鼠的 aPFC（前边缘和下边缘皮质）会增加不耐烦的尝试并减少等待时间（Murakami 等人，2017 年）。有关前额叶区域在行为抑制中的作用的详细评论，请参阅 Kamigaki, 2019。另请参阅 MG White 等人，2018。
patient to get food: Inactivating prelimbic area of rat frontal cortex (part of their aPFC) increases premature responses (e.g., lever-release before go stimulus). (Hardung et al., 2017; Narayanan et al., 2006). Inactivation of the aPFC (prelimbic and infralimbic cortex) in rats increases impatient trials and decreases waiting time (Murakami et al., 2017). For a good review on the role of prefrontal areas in behavioral inhibition, see Kamigaki, 2019. Also see M. G. White et al., 2018.
朝着想象的目标前进： Procyk 等人，2000 年；Procyk 和 Joseph，2001 年。后者观察到猴子前扣带皮层（aPFC 的一部分）中的神经元对按顺序执行的动作的顺序很敏感（即使实际执行的动作相同），这表明大脑的这个区域正在模拟动物所处的整体序列，而不仅仅是单个动作。在大鼠（Cowen 等人，2012 年；Cowen 和 McNaughton，2007 年）和人类（Koechlin 等人，2002 年）中也有这种证据。
toward an imagined goal: Procyk et al., 2000; Procyk and Joseph, 2001. The latter observed that neurons in the monkey anterior cingulate cortex (a part of aPFC) are sensitive to the order of actions executed in a sequence (even if the actual movements performed are the same), suggesting that this area of the brain is modeling the overall sequence an animal is within, not just individual movements. There is also evidence for this in rats (Cowen et al., 2012; Cowen and McNaughton, 2007) and humans (Koechlin et al., 2002).
在一项正在进行的任务中： Dehaene 等人，1994 年；MacDonald 等人，2000 年；Ridderinkhof 等人，2004 年；Totah 等人，2009 年。有关有趣的综合研究，另请参阅 Shenhav 等人，2016 年。
in an ongoing task: Dehaene et al., 1994; MacDonald et al., 2000; Ridderinkhof et al., 2004; Totah et al., 2009. For an interesting synthesis, also see Shenhav et al., 2016.
据报道为：有关此内容的进一步阅读，请参阅 Gal，2016 年和 Lakshminarayanan 等人，2017 年。
is reported to be: For some further reading on this, see Gal, 2016, and Lakshminarayanan et al., 2017.
基底神经节：额叶皮质向基底神经节的一部分（称为丘脑底核）发送直接投射，这已被证明能够完全停止行为（Narayanan 等人，2020 年）。
of the basal ganglia: The frontal cortex sends a direct projection to a part of the basal ganglia called the subthalamic nucleus that has been shown to be able to completely halt behavior (Narayanan et al., 2020).
具有不确定性水平： ER Stern 等人，2010 年。
with levels of uncertainty: E. R. Stern et al., 2010.
感觉新皮质的活动：有关该回路的详细评论，请参阅 Kamigaki, 2019。请注意，不同的额叶区域可能与不同的感觉模式相关 - aPFC 的某些子区域（如前扣带皮层）主要将输出发送到视觉区域而不是躯体感觉和听觉区域，而其他子区域主要将输出发送到听觉和躯体感觉区域（S. Zhang 等人，2016 年）。
activity of the sensory neocortex: For a good review of this circuitry, see Kamigaki, 2019. Note that different frontal regions may be associated with different sensory modalities—some subregions of the aPFC (like the anterior cingulate cortex) mostly send output to visual and not somatosensory and auditory areas, whereas others mostly send output to auditory and somatosensory areas (S. Zhang et al., 2016).
皮质变得独特同步： Benchenane 等人，2010 年；Spellman 等人，2015 年，表明在替代性试错过程中，aPFC 和海马体之间存在同步，如 Redish，2016 年所评论的。Hyman 等人，2010 年；MW Jones 和 Wilson，2005 年，表明在情景记忆任务期间，海马体和 PFC 之间存在同步。Sauseng 等人，2004 年；Sederberg 等人，2003 年；Xie 等人，2021 年，表明在工作记忆和情景记忆任务期间，前额叶和感觉新皮质之间存在同样的同步。
cortex become uniquely synchronized: Benchenane et al., 2010; Spellman et al., 2015, show synchronization between the aPFC and hippocampus during vicarious trial and error, as reviewed in Redish, 2016. Hyman et al., 2010; M. W. Jones and Wilson, 2005, show synchronization between the hippocampus and PFC during episodic memory tasks. Sauseng et al., 2004; Sederberg et al., 2003; Xie et al., 2021, show this same synchronization between prefrontal and sensory neocortex during working memory and episodic-memory tasks.
选择一个动作： Bogacz 和 Gurney，2007 年； Krajbich 等人，2010。
an action is selected: Bogacz and Gurney, 2007; Krajbich et al., 2010.
那些没有的： Dickinson，1985。
those that had not: Dickinson, 1985.
食物贬值了：同上。
the food was devalued: Ibid.
称之为“主动推理”： Adams 等人，2013 年。
calls this “active inference”: Adams et al., 2013.
缺少第四层：他的理论是关于运动皮层的，这也是无颗粒的（见下一章），但逻辑同样适用于 aPFC。
missing the fourth layer: His theory was about the motor cortex, which is also agranular (see next chapter), but the logic applies equally to aPFC.
（仰望天空）：有证据表明，当不同层被抑制时，新皮质柱会在不同状态之间振荡，当你想象和关注事物时，它们会以不同的节奏振荡。我在 Bennett，2020 年回顾了这些证据。
(looking at the sky): There is some evidence that neocortical columns oscillate between different states when different layers are suppressed and they oscillate at different rhythms when you’re imagining versus when you’re attending to things. I review the evidence in Bennett, 2020.
鸭子或兔子： S. Zhang 等人，2014 年，展示了用于调节感觉新皮层中表征的额叶回路。
duck or a rabbit: S. Zhang et al., 2014, shows frontal circuits for modulating representations in the sensory neocortex.
蜥蜴数百次试验：瓦格纳，1932年。
lizards hundreds of trials: Wagner, 1932.
如果你损坏了老鼠的 aPFC： Dias 和 Aggleton，2000 年。请注意，老鼠仍然可以以大约正常的速度学习非匹配位置任务，因为它们不必克服本能倾向，避免最近利用的觅食地点（避免它们刚刚体验过的地方是本能）。Passingham 和 Wise，2015 年对此进行了讨论。
if you damage a rat’s aPFC: Dias and Aggleton, 2000. Note that rats can still learn non-matching-to-position tasks at approximately normal rates because they do not have to overcome their instinctual tendency to avoid recently exploited foraging sites (avoiding the place they just experienced is instinctual). Discussed in Passingham and Wise, 2015.
第 14 章：洗碗机器人的秘密
Chapter 14: The Secret to Dishwashing Robots
皮质损伤： Darling 等人，2011 年。
cortex damage: Darling et al., 2011.
大脑皮层的其他区域：但请注意，运动皮层是“无颗粒的”，这意味着它具有薄的或者缺失的第四层（就像无颗粒的前额叶皮层一样）。
other areas of the neocortex: Although note that the motor cortex is “agranular,” meaning it has a thin or missing layer four (just like agranular prefrontal cortex).
运动皮层的进化： Karlen 和 Krubitzer，2007 年。
evolution of the motor cortex: Karlen and Krubitzer, 2007.
蝙蝠、大象和猫： Karlen 和 Krubitzer，2007 年。胎盘动物和有袋动物的分化被认为发生于大约 1.6 亿年前的侏罗纪时期（ZX Luo 等人，2011 年）。
bats, elephants, and cats: Karlen and Krubitzer, 2007. The placental-marsupial divergence is believed to have occurred about 160 million years ago in the Jurassic period (Z. X. Luo et al., 2011).
患有这种瘫痪： Kawai 等人，2015；Whishaw 等人，1991。灵长类动物运动皮层中的神经元绕过较旧的回路并与脊髓神经元直接连接（Lemon，2019）。尽管一些证据表明这种直接投射也可能发生在大鼠中（Elger 等人，1977；Gu 等人，2017；Maeda 等人，2016），但新的证据表明，这些直接投射在成年期会消失（Murabe 等人，2018），这与灵长类动物不同（Armand 等人，1997；Eyre，2007）。灵长类动物运动皮层损伤的瘫痪似乎并不代表早期哺乳动物的运动皮层。
suffer from such paralysis: Kawai et al., 2015; Whishaw et al., 1991. Neurons in the motor cortex of primates bypass older circuits and make direct connections with spinal neurons (Lemon, 2019). Although some evidence suggests such direct projections may also occur in rats (Elger et al., 1977; Gu et al., 2017; Maeda et al., 2016), new evidence shows that these direct projections disappear in adulthood (Murabe et al., 2018), unlike in primates (Armand et al., 1997; Eyre, 2007). The paralysis of motor-cortex damage in primates does not seem to be representative of the motor cortex in early mammals.
小型不平整平台： Alaverdashvili 和 Whishaw，2008 年；T. Drew 等人，2008 年；T. Drew 和 Marigold，2015 年；Grillner 和 el Manira，2020 年。
small unevenly placed platform: Alaverdashvili and Whishaw, 2008; T. Drew et al., 2008; T. Drew and Marigold, 2015; Grillner and el Manira, 2020.
学习杠杆序列： Kawai 等人，2015 年。
learn the lever sequence: Kawai et al., 2015.
图 14.3： Rebecca Gelernter 的艺术作品；这个特殊的人物形象受到了 Grillner 和 el Manira（2020 年）图像的启发。
Figure 14.3: Art by Rebecca Gelernter; this particular figure was inspired by imagery from Grillner and el Manira, 2020.
需要规划的运动： Beloozerova 等人，2010 年；Farrell 等人，2014 年；Grillner 和 el Manira，2020 年。
movements that require planning: Beloozerova et al., 2010; Farrell et al., 2014; Grillner and el Manira, 2020.
已知存在： Andujar 等人，2010 年；Beloozerova 和 Sirota，2003 年；T. Drew 和 Marigold，2015 年。
known to be there: Andujar et al., 2010; Beloozerova and Sirota, 2003; T. Drew and Marigold, 2015.
它大概计划的运动： Lajoie 等人，2010 年。
movement it presumably planned: Lajoie et al., 2010.
运动皮层被激活： Malouin 等人，2003 年。
motor cortex becomes activated: Malouin et al., 2003.
与运动无关： Kosonogov，2011。
unrelated to movement: Kosonogov, 2011.
甚至手术操作： Arora 等人，2011 年。
and even surgical maneuvers: Arora et al., 2011.
整个事件是： Kohlsdorf 和 Navas，2007 年；Olberding 等人，2012 年；Parker 和 McBrayer，2016 年；Tucker 和 McBrayer，2012 年。
the whole affair is: Kohlsdorf and Navas, 2007; Olberding et al., 2012; Parker and McBrayer, 2016; Tucker and McBrayer, 2012.
绕过平台： Kohlsdorf 和 Biewener，2006 年；奥尔伯丁等人，2012；自我，2012。
to get around platforms: Kohlsdorf and Biewener, 2006; Olberding et al., 2012; Self, 2012.
可以治愈药物成瘾： N. Li 等人，2013 年。
can cure drug addiction: N. Li et al., 2013.
低级子目标： Lashley，1951；Yokoi 和 Diedrichsen，2019。
lower-level subgoals: Lashley, 1951; Yokoi and Diedrichsen, 2019.
层次结构中的级别： Thorn 等人，2010 年。
levels in the hierarchy: Thorn et al., 2010.
（他们经历的考验）： Yin 等人，2004 年。
(trials they go through): Yin et al., 2004.
不受其控制：Brainin 等人，2008 年。
own without their control: Brainin et al., 2008.
运动前皮层损伤： P. Gao 等，2003 年。
damage to the premotor cortex: P. Gao et al., 2003.
附近的刺激暗示： Lhermitte，1983。
that nearby stimuli suggest: Lhermitte, 1983.
率下降至42%： N. Li et al.，2013。
rate dropped to 42 percent: N. Li et al., 2013.
第 15 章：政治智慧的军备竞赛
Chapter 15: The Arms Race for Political Savvy
陆生脊椎动物： Sahney 和 Benton，2008 年。
of land-living vertebrates: Sahney and Benton, 2008.
图 15.2：图自同上。
Figure 15.2: Figure from ibid.
更大的社会群体： Dunbar，1998。
bigger its social group: Dunbar, 1998.
对于大多数其他动物： Pérez-Barbería 等人，2007 年；Shultz 和 Dunbar，2007 年。Dunbar 和 Shultz，2017 年审阅。
for most other animals: Pérez-Barbería et al., 2007; Shultz and Dunbar, 2007. Reviewed in Dunbar and Shultz, 2017.
低下头并移开视线： Stringham，2011 年。
down and look away: Stringham, 2011.
并使耳朵变平： S.Curtis，1998。
and flatten their ears: S. Curtis, 1998.
多雄性群体： Shultz 和 Dunbar，2007 年。原始数据来自 Nowak，1999 年。请注意，这些分类并不准确，并非所有亚种都属于其中一种，也可能有其他分类。但这些是常见的，也是不同类型社会组织的大致近似值。Dunbar 在他的一些开创性著作中使用了这四种分类，它们是灵长类文献中的标准（BB Smuts 等人，1987 年）。
multi-male groups: Shultz and Dunbar, 2007. Original data from Nowak, 1999. Note these classifications are inexact, not all subspecies fall into one or the other, and there may be other categorizations as well. But these are the common ones and are the broad first approximations of different types of social organizations. Dunbar used these four categorizations in some of his seminal work, and they are standard in primate literature (B. B. Smuts et al., 1987).
哺乳动物中发现的四种常见社会结构： Shultz 和 Dunbar，2007 年。原始数据来自 Nowak，1999 年。
four common social structures found in mammals: Shultz and Dunbar, 2007. Original data from Nowak, 1999.
自己的孩子：虽然大多数组织“后宫”的哺乳动物都有一只雄性和多只雌性，但确实存在角色互换的情况。一些狨猴、猴子和有袋动物有一只雌性，它们与多只雄性交配（Goldizen，1988）。当然，在许多昆虫中，比如蜜蜂，由雌性管理的社会群体并不是例外，而是常态。
his own children: While most harem-organizing mammals have a single male and multiple females, there are indeed cases where the roles are reversed. Some marmosets, monkeys, and marsupials have single females that mate with multiple males (Goldizen, 1988). And of course, in many insects, like bees, female-run social groups are less the exception and more the rule.
避免大型团体活动： RA Hill 和 Dunbar，1998 年。
avoidance of large groups: R. A. Hill and Dunbar, 1998.
他们创建了回应： Bettridge 等人，2010 年。
they created in response: Bettridge et al., 2010.
生活在一英亩的森林里：门泽尔，1974 年。故事总结于柯克帕特里克，2007 年。
living in a one-acre forest: Menzel, 1974. Story summarized in Kirkpatrick, 2007.
“意外”和“故意”行为： Call 和 Tomasello，1998 年。
“accidental” and “intentional” actions: Call and Tomasello, 1998.
那些看起来不愿意的人： Call 等人，2004 年。
those who seemed unwilling: Call et al., 2004.
护目镜看不到它们： Kano 等人，2019 年。
goggles wouldn’t see them: Kano et al., 2019.
像一些鸟类、海豚： Tomonaga，2010。
like some birds, dolphins: Tomonaga, 2010.
培训师了解的地点： Bräuer，2014；卡明斯基等人，2009。
location the trainer knows about: Bräuer, 2014; Kaminski et al., 2009.
梳理和群体规模：Dunbar，1991，1998；Lehmann 等人，2007。
grooming and group size: Dunbar, 1991, 1998; Lehmann et al., 2007.
通过更多地打扮自己： RM Seyfarth，1980年。
by grooming themselves more: R. M. Seyfarth, 1980.
通过外表和声音：斯诺登和克利夫兰，1980 年。
by appearance and voice: Snowdon and Cleveland, 1980.
母亲的作用： DL Cheney 和 Seyfarth，1980 年，1982 年。
what the mother does: D. L. Cheney and Seyfarth, 1980, 1982.
A 将提交给 C： Andelman，1985；RM Seyfarth，1980。
A will submit to C: Andelman, 1985; R. M. Seyfarth, 1980.
很多年，甚至几代人： D. Cheney，1983 年。
many years, even generations: D. Cheney, 1983.
播放这些录音： Bergman 等人，2003 年。
play back these recordings: Bergman et al., 2003.
等等： Berman，1982；Horrocks 和 Hunte，1983；Walters 和 Seyfarth，1987。
on and so forth: Berman, 1982; Horrocks and Hunte, 1983; Walters and Seyfarth, 1987.
她母亲的排名： Berman，1983年；Lee，1983年。
rank of her mother: Berman, 1983; Lee, 1983.
较高地位家族： Datta，1983 年。
of higher-ranking families: Datta, 1983.
死于疾病： Silk，1987年；Silk 等人，2003年、2010年。
to die from disease: Silk, 1987; Silk et al., 2003, 2010.
层次结构已经建立： MRA Chance 等，1977 年；Gouzoules，1980 年。
hierarchy has been established: M. R. A. Chance et al., 1977; Gouzoules, 1980.
非家庭成员：在 DL Cheney 和 Seyfarth（2019 年）第 2 章中进行了评论。
nonfamily members that come: Reviewed in chapter 2 of D. L. Cheney and Seyfarth, 2019.
招募这样的盟友： Chapais，1988 年；也在 DL Cheney 和 Seyfarth（2019 年）第 2 章中讨论。
to recruit such allies: Chapais, 1988; also discussed in chapter 2 of D. L. Cheney and Seyfarth, 2019.
与以下机构建立了美容合作伙伴关系： D. Cheney，1983 年。
formed grooming partnerships with: D. Cheney, 1983.
发出“救救我”的声音： RM Seyfarth 和 Cheney，1984 年。
makes a “help me” vocalization: R. M. Seyfarth and Cheney, 1984.
为自己辩护： FBM de Waal，1982 年；Packer，1977 年；BB Smuts，2017 年。
to their own defense: F. B. M. de Waal, 1982; Packer, 1977; B. B. Smuts, 2017.
只为自己提供食物： Engelmann 和 Herrmann，2016 年。
food just for themselves: Engelmann and Herrmann, 2016.
与基思打交道：Datta，1983；Dunbar，2012。
to deal with Keith: Datta, 1983; Dunbar, 2012.
获取更多食物的途径： Packer，1977；Silk，1982。
more access to food: Packer, 1977; Silk, 1982.
排名高于自己： Cheney 和 Seyfarth，2019 年。
ranked higher than themselves: Cheney and Seyfarth, 2019.
该小组成员： Gouzoules，1975 年；Scott，1984 年。
members of the group: Gouzoules, 1975; Scott, 1984.
与高级个人： DL Cheney 和 Seyfarth，2019 年。
with high-ranking individuals: D. L. Cheney and Seyfarth, 2019.
级别较高的个体： D. Cheney, 1983 年；DL Cheney, 1977 年；RM Seyfarth, 1977 年。
the higher-ranking individual: D. Cheney, 1983; D. L. Cheney, 1977; R. M. Seyfarth, 1977.
最受欢迎的玩伴： P. Lee，1983年。
the most popular playmates: P. Lee, 1983.
你在我的羽翼下：Stammbach，1988b，1988a。
you under my wing: Stammbach, 1988b, 1988a.
与非家庭成员： Cheney 和 Seyfarth，1989 年。
with nonfamily members: Cheney and Seyfarth, 1989.
最近争吵过： Cheney 和 Seyfarth，2019 年。
have recently quarreled with: Cheney and Seyfarth, 2019.
进食、休息和交配： Dunbar 等人，2009 年。
eating, resting, and mating: Dunbar et al., 2009.
比大多数其他哺乳动物： Dunbar，1991。
than most other mammals: Dunbar, 1991.
花更多时间社交： Borgeaud 等人，2021 年。
spend more time socializing: Borgeaud et al., 2021.
与社交敏锐度： Byrne and Corp，2004。
with social savviness: Byrne and Corp, 2004.
第 16 章：如何模仿他人的思维
Chapter 16: How to Model Other Minds
约三百五十克： Ginneken 等人，2017 年；托比亚斯，1971；范埃森等人，2019。
about three hundred fifty grams: Ginneken et al., 2017; Tobias, 1971; van Essen et al., 2019.
基本方式相同：公平地说，灵长类动物的新皮质中神经元的密度确实更高，但这种更密集的排列并不意味着新皮质柱的整体结构发生了变化，只是它被放大并挤进了一个更小的区域。有关大脑缩放的详细评论，请参阅 Herculano-Houzel，2012 年。
the same fundamental ways: To be fair, primates do have a higher density of neurons in their neocortices, but this denser packing doesn’t suggest a change to the overall architecture of a neocortical column, merely that it has been scaled up and packed into a smaller area. For a good review of brain scaling, see Herculano-Houzel, 2012.
额叶皮质的补充： Preuss，2009。
addition to the frontal cortex: Preuss, 2009.
和感觉新皮质区域： Goldman-Rakic，1988；Gutierrez 等人，2000。
and sensory neocortical regions: Goldman-Rakic, 1988; Gutierrez et al., 2000.
根本的功能意义： Hebb，1945；赫布和彭菲尔德，1940； HL Teuber 和韦恩斯坦，1954 年。
functional significance at all: Hebb, 1945; Hebb and Penfield, 1940; H. L. Teuber and Weinstein, 1954.
智力或感知： Hebb 和 Penfield，1940 年。
intellect or perception whatsoever: Hebb and Penfield, 1940.
皮层是一个“谜”： H. Teuber，1964年。
cortex was a “riddle”: H. Teuber, 1964.
gPFC 亮起： Gusnard 等人，2001 年。
gPFC lit up: Gusnard et al., 2001.
关于您自己的一般情况： Christoff 等人，2009 年；Herwig 等人，2010 年；Kelley 等人，2002 年；Moran 等人，2006 年；Northoff 等人，2006 年；Schmitz 等人，2004 年。
about yourself in general: Christoff et al., 2009; Herwig et al., 2010; Kelley et al., 2002; Moran et al., 2006; Northoff et al., 2006; Schmitz et al., 2004.
周围元素： Kurczek 等人，2015 年。
of the surrounding elements: Kurczek et al., 2015.
镜子中的自己： Breen 等人，2001 年；Postal，2005 年；Spangenberg 等人，1998 年。
themselves in a mirror: Breen et al., 2001; Postal, 2005; Spangenberg et al., 1998.
杏仁核和海马体： Morecraft 等人，2007 年；Insausti 和 Muñoz，2001 年。
the amygdala and hippocampus: Morecraft et al., 2007; Insausti and Muñoz, 2001.
来自较老的 aPFC： Ray 和 Price，1993 年。进一步的证据是，刺激灵长类动物的无颗粒皮质会引起自主神经效应（呼吸频率、血压、脉搏、瞳孔扩张和竖毛的变化），而在颗粒皮质中则不会（Kaada，1960 年；Kaada 等人，1949 年）。
from the older aPFC: Ray and Price, 1993. Further evidence for this is seen in the fact that stimulation of the agranular cortex in primates elicits autonomic effects (changes in respiratory rate, blood pressure, pulse, pupillary dilation, and piloerection), while in the granular cortex, it does not (Kaada, 1960; Kaada et al., 1949).
激活后点亮： Brunet 等人，2000 年；Völlm 等人，2006 年。连环画任务类似于 Baron-Cohen 等人（1986 年）早期的讲故事工作。
lit up with activation: Brunet et al., 2000; Völlm et al., 2006. The comic-strip task is similar to the earlier storytelling work by Baron-Cohen et al., 1986.
图 16.4：图像来自 Brunet 等人，2000 年；Völlm 等人，2006 年；以及与 Eric Brunet-Gouet 博士的私人通信。经 Brunet-Gouet 博士许可使用（私人通信）。
Figure 16.4: Images from Brunet et al., 2000; Völlm et al., 2006; and personal correspondence with Dr. Eric Brunet-Gouet. Used with permission of Dr. Brunet-Gouet (personal correspondence).
四岁： HM Wellman 等，2001 年；H. Wimmer 和 Perner，1983 年。
the age of four: H. M. Wellman et al., 2001; H. Wimmer and Perner, 1983.
图 16.5： Frith，2003 年的照片。已获得许可重复使用。
Figure 16.5: Photo from Frith, 2003. Reused with permission.
激活程度： Gweon et al., 2012; Otsuka et al., 2009; Saxe and Kanwisher, 2003; Young et al., 2007。
the degree of activation: Gweon et al., 2012; Otsuka et al., 2009; Saxe and Kanwisher, 2003; Young et al., 2007.
此类错误信念任务： Carrington 和 Bailey，2009 年；van Overwalle 和 Baetens，2009 年，特别指出颗粒前额叶皮质的两个区域（背内侧前额叶皮质和前内侧前额叶皮质，大致构成布罗德曼第 8、9 和 10 区）以及颞顶交界处和颞上沟是仅由需要心理理论的任务激活的区域。
such false-belief tasks: Carrington and Bailey, 2009; van Overwalle and Baetens, 2009, implicate specifically two areas of the granular prefrontal cortex (dorsomedial prefrontal cortex and anteromedial prefrontal cortex, which roughly make up Brodmann areas 8, 9, and 10) as well as the temporoparietal junction and the superior temporal sulcus as areas that are uniquely activated by tasks that require the theory of mind.
Sally-Ann 检验： Siegal 等人，1996 年；VE Stone 等人，1998 年。
the Sally-Ann test: Siegal et al., 1996; V. E. Stone et al., 1998.
其他人的情绪： Shaw 等人，2005 年。
emotions in other people: Shaw et al., 2005.
其他人的情绪： Shamay-Tsoory 等人，2003 年。
with other people’s emotions: Shamay-Tsoory et al., 2003.
区分谎言和笑话： Winner 等人，1998 年。
distinguish lies from jokes: Winner et al., 1998.
可能会冒犯某些人： Shamay-Tsoory 等人，2005 年；VE Stone 等人，1998 年。
that would offend someone: Shamay-Tsoory et al., 2005; V. E. Stone et al., 1998.
别人的视觉视角： Stuss et al., 2001。
someone else’s visual perspective: Stuss et al., 2001.
欺骗别人的斗争：同上。
struggle to deceive others: Ibid.
对非人类灵长类动物的影响： Dehaene 等人，2005 年；DI Perrett 等人，1992 年；Ramezanpour 和 Thier，2020 年。
effects in nonhuman primates: Dehaene et al., 2005; D. I. Perrett et al., 1992; Ramezanpour and Thier, 2020.
在人类中确实如此： T. Hayashi 等人，2020 年。
it does in humans: T. Hayashi et al., 2020.
灵长类动物的社交网络规模： Sallet 等人，2011 年。
social-network size in primates: Sallet et al., 2011.
心智理论任务： J. Powell 等人，2012 年；Stiller 和 Dunbar，2007 年；PA Lewis 等人，2011 年；JL Powell 等人，2010 年。
theory-of-mind tasks: J. Powell et al., 2012; Stiller and Dunbar, 2007; P. A. Lewis et al., 2011; J. L. Powell et al., 2010.
模仿他人的思想：请参阅 Amodio 和 Frith（2006 年）关于前额叶皮层中与自我参照和思考他人有关的特定区域的详细评论。
to model other minds: See Amodio and Frith, 2006, for a detailed review of the specific areas in the prefrontal cortex implicated in self-reference and thinking about others.
或“社会投射理论”： Gallese 和 Goldman，1998 年；Goldman，1992 年；Gordon，2011 年；Harris，1992 年。需要注意的是，并非所有人都同意这些是通过同一过程实现的。有关这场辩论的一些好的评论，请参阅 Dimaggio 等人，2008 年；Gallup，1998 年。
or “social projection theory”: Gallese and Goldman, 1998; Goldman, 1992; Gordon, 2011; Harris, 1992. It should be noted that not everyone agrees that these are implemented by the same process. For some good reviews of this debate, see Dimaggio et al., 2008; Gallup, 1998.
相同的过程：当评估自己的性格特征或接受他人对自己的评价时，gPFC 中的相同心理化网络会被激活 - 具体来说，是前额叶皮质的内侧区域（Ochsner 等人，2005 年）。
the same process: When evaluating your own personality traits or receiving evaluations of yourself by others, the same mentalizing network in the gPFC activates—specifically, the medial area of the prefrontal cortex (Ochsner et al., 2005).
心智理论：与此观点相一致的是，他人的心智理论建立在自我生成模型之上，而自我概念在儿童发展过程中的出现早于心智理论的出现。参见 Keenan 等人，2005 年；Ritblatt，2000 年；Rochat，1998 年。
of theory of mind: Further consistent with this idea that theory of mind of others is bootstrapped on a generative model of yourself is the fact that the concept of self emerges in child development before the theory of mind emerges. See Keenan et al., 2005; Ritblatt, 2000; Rochat, 1998.
大约两岁：阿姆斯特丹，1972 年。
about two years old: Amsterdam, 1972.
想要、希望和假装： Frith and Frith，2003 年。
want, wish, and pretend: Frith and Frith, 2003.
“知道它是鳄鱼”： Shatz 等人，1983 年。
“know it’s a crocodile”: Shatz et al., 1983.
尊重其他人： HM Wellman 等人，2001 年。
with respect to other people: H. M. Wellman et al., 2001.
另一方则表现更好： Gopnik 和 Meltzoff，2011 年；Lang 和 Perner，2002 年。
better at the other: Gopnik and Meltzoff, 2011; Lang and Perner, 2002.
在镜子中认出自己： Gallup 等人，1971 年。有证据表明大象和海豚可以在镜子中认出自己（Plotnik 等人，2006 年；Reiss 和 Marino，2001 年）。猿类，如黑猩猩、猩猩和大猩猩，也具有识别能力（Suarez 和 Gallup，1981 年；Posada 和 Colell，2007 年）。猴子也可能在镜子中认出自己（Chang 等人，2017 年）。有关这些镜子测试的精彩总结，请参阅 Suddendorf，2013 年第 3 章。
themselves in a mirror: Gallup et al., 1971. There is some evidence suggesting that elephants and dolphins can recognize themselves in a mirror (Plotnik et al., 2006; Reiss and Marino, 2001). Recognition is seen in apes, such as chimps, orangutans, and gorillas (Suarez and Gallup, 1981; Posada and Colell, 2007). Monkeys may also recognize themselves in a mirror (Chang et al., 2017). A great summary of these mirror tests can be found in chapter 3 of Suddendorf, 2013.
其他人的这些状态： Kawada et al., 2004;尼登塔尔等人，2000。
these states in others: Kawada et al., 2004; Niedenthal et al., 2000.
比它们更口渴： van Boven 和 Loewenstein，2003 年。
thirstier than they are: van Boven and Loewenstein, 2003.
人格特质对他人的影响： Bargh 和 Chartrand，2000 年。
personality traits onto others: Bargh and Chartrand, 2000.
他人的思想：迈克尔·格拉齐亚诺 (Michael Graziano) 对此及其与意识的关系有一些有趣的想法。他认为，我们的祖先进化出了心理理论来指导他们独特的社会生活，而这种理论的一个副作用是，当他们将这种心理理论应用于内心时，意识就出现了 (Graziano, 2019)。
the minds of others: Michael Graziano has some fascinating ideas about this and its relationship to consciousness. He argues that our ancestors evolved theory of mind to navigate their unique social lives and a side effect of this was that when they applied this theory of mind inward, consciousness emerged (Graziano, 2019).
(“这是跳跃”）： Surís 等人，2021 年。
(“This is jumping”): Surís et al., 2021.
（按情绪分类的面孔）：克劳福德，2021 年。
(faces classified by emotions): Crawford, 2021.
第 17 章：猴子锤和自动驾驶汽车
Chapter 17: Monkey Hammers and Self-Driving Cars
清洁耳朵：使用牙线：Pal 等人，2018 年。不同技术列表：Sanz 和 Morgan，2007 年。
to clean their ears: Flossing: Pal et al., 2018. Lists of different techniques: Sanz and Morgan, 2007.
苍蝇和抓挠自己： Hart 等人，2001 年。
flies and scratch themselves: Hart et al., 2001.
打开坚果： Müller，2010 年。
to break open nuts: Müller, 2010.
内在食物： Bernardi，2012年。
to the inner food: Bernardi, 2012.
不同的工具使用行为： Sanz 和 Morgan，2007。
different tool-using behaviors: Sanz and Morgan, 2007.
比贡贝的要多： Musgrave 等人，2020 年。
than those in Gombe: Musgrave et al., 2020.
及其他用于撕裂的材料： di Pellegrino 等，1992 年。
and others for tearing: di Pellegrino et al., 1992.
刚刚发生：故事在 Taylor（2016 年）和 Roche 和 Commins（2009 年）中讲述。
had just happened: Story told in Taylor, 2016 and Roche and Commins, 2009.
（伸出舌头）： di Pellegrino 等，1992；法拉利等人，2003；加莱塞等人，1996。
(sticking one’s tongue out): di Pellegrino et al., 1992; Ferrari et al., 2003; Gallese et al., 1996.
（顶叶，运动皮层）： di Pellegrino 等，1992；杜沙诺瓦和多诺霍，2010；福加西等人，2005；特卡奇等人，2007。
(parietal lobe, motor cortex): di Pellegrino et al., 1992; Dushanova and Donoghue, 2010; Fogassi et al., 2005; Tkach et al., 2007.
许多灵长类动物： Brass 等人，2007 年；Buccino 等人，2001 年；Mukamel 等人，2010 年。
numerous species of primates: Brass et al., 2007; Buccino et al., 2001; Mukamel et al., 2010.
镜像神经元的解释：有关镜像神经元当前争论的评论，请参阅 Heyes 和 Catmur，2022 年；Hickok，2014 年；Jeon 和 Lee，2018 年；Rozzi，2015 年。
interpretations of mirror neurons: For reviews of the current debate around mirror neurons, see Heyes and Catmur, 2022; Hickok, 2014; Jeon and Lee, 2018; Rozzi, 2015.
心智理论： Rizzolatti 等人，2001 年。
in theory of mind: Rizzolatti et al., 2001.
看看别人怎么做： Gallese 和 Goldman，1998 年。
see someone else do: Gallese and Goldman, 1998.
打开（什么都看不见）： Kohler 等人，2002 年。
open (without seeing anything): Kohler et al., 2002.
（墙后的盒子）： Umiltà 等人，2001 年。
(box behind the wall): Umiltà et al., 2001.
（短语，如梳理头发）： Pazzaglia et al., 2008；塔尔汉等人，2015；乌尔杰西等人，2014。
(phrase, like combing hair): Pazzaglia et al., 2008; Tarhan et al., 2015; Urgesi et al., 2014.
自行弹跳： Pobric 和 Hamilton，2006 年。与此一致，如果人们主动拿起一个轻便的盒子，他们就会倾向于认为他们看到其他人拿起的盒子也很轻。当一个人真正拿起盒子而不是被动地拿着一个轻便的盒子时，这种偏见会大得多，这表明这并不是将盒子与轻便联系起来，而是关于自己拿起盒子的主动体验（A. Hamilton 等人，2004 年）。
bouncing on its own: Pobric and Hamilton, 2006. Consistent with this, if people actively pick up a light box, they become biased toward thinking that boxes they see others pick up are also light. This bias is far greater when an individual actually lifts the box as opposed to passively holding a light box, which demonstrates it isn’t about associating a box with lightness but about the active experience of picking up a box yourself (A. Hamilton et al., 2004).
（吃汉堡，吹灭蜡烛）： Michael 等人，2014 年。
(eating a burger, blowing out a candle): Michael et al., 2014.
意图成立： Thompson 等人（2019 年）对其中一些想法进行了很好的回顾。但请参阅 Negri 等人（2007 年）和 Vannuscorps 和 Caramazza（2016 年）的反例，这些反例表明，行动产生的障碍并不总是会损害行动感知。
intended to hold: Thompson et al., 2019, provides a nice review of some of these ideas. But see Negri et al., 2007 and Vannuscorps and Caramazza, 2016 for counterexamples that suggest that impairments to action production don’t always impair action perception.
变得更加活跃： S. Vogt 等人，2007 年。
becomes way more activated: S. Vogt et al., 2007.
按照红点： Catmur 等人，2009 年；Heiser 等人，2003 年。
following the red dots: Catmur et al., 2009; Heiser et al., 2003.
学会这样做： Humle 等人，2009 年；Lonsdorf，2005 年。
learns to do it: Humle et al., 2009; Lonsdorf, 2005.
晚年技能： Biro et al., 2003；松泽等人，2008。
skill later in life: Biro et al., 2003; Matsuzawa et al., 2008.
（笼子里的食物）： M. Hayashi 等人，2005 年；马歇尔-佩西尼和怀特恩，2008；托马塞洛等人，1987；苏比乌尔等人，2004。
(food in the cage): M. Hayashi et al., 2005; Marshall-Pescini and Whiten, 2008; Tomasello et al., 1987; Subiaul et al., 2004.
以正确的方式： Whiten 等人，2005 年。
in the right way: Whiten et al., 2005.
获取食物的方式： Dindo 等人，2009 年。
way to get food: Dindo et al., 2009.
抽屉获取食物： Gunhold 等人，2014 年。
drawer to get food: Gunhold et al., 2014.
打开人造水果： E. van de Waal 等，2015。
opening an artificial fruit: E. van de Waal et al., 2015.
传承多代： Haslam 等人，2016 年；Mercader 等人，2007 年；Whiten，2017 年。
down through multiple generations: Haslam et al., 2016; Mercader et al., 2007; Whiten, 2017.
杠杆并取水： Zentall 和 Levine，1972 年。
lever and get water: Zentall and Levine, 1972.
其父母的技术： Müller 和 Cant，2010。
technique of their parents: Müller and Cant, 2010.
其他海豚或人类： Hermann，2002。
other dolphins or humans: Hermann, 2002.
狗表演该动作： Range 等人，2007 年。
dog perform the act: Range et al., 2007.
那些相同的导航路径：有关鱼类的观察学习，请参阅 Lindeyer 和 Reader，2010 年。有关爬行动物，请参阅 Kis 等人，2015 年；Wilkinson，Kuenstner 等人，2010 年；Wilkinson，Mandl 等人，2010 年。
those same navigational paths: For observational learning in fish, see Lindeyer and Reader, 2010. For reptiles, see Kis et al., 2015; Wilkinson, Kuenstner, et al., 2010; Wilkinson, Mandl, et al., 2010.
积极地互相教导：有关这方面的论据，请参阅 Hoppitt 等人，2008 年；Kline，2014 年；Premack，2007 年。
actively teach one another: For arguments on this, see Hoppitt et al., 2008; Kline, 2014; Premack, 2007.
年轻人的手： Boesch，1991。
hands of their young: Boesch, 1991.
下来帮助教学： Masataka 等人，2009 年。
down to help teach: Masataka et al., 2009.
与他们交换工具： Musgrave 等人，2016 年。
swap tools with them: Musgrave et al., 2016.
给年轻人的工具：Musgrave 等人，2020 年。
tool to a youngster: Musgrave et al., 2020.
跳过了不相关的步骤： Call 等，2005；Horner 和 Whiten，2005；Nagell 等，1993。
skipped the irrelevant steps: Call et al., 2005; Horner and Whiten, 2005; Nagell et al., 1993.
几圈：故事讲述于克里斯蒂安的《对齐问题》，232。
a handful of laps: Story told in Christian, The Alignment Problem, 232.
称为“逆向强化学习”： Abbeel、Coates 和 Ng，2004 年。
called “inverse reinforcement learning”: Abbeel, Coates, and Ng, 2004.
遥控直升机： Abbeel 等人，2010 年。
a remote-controlled helicopter: Abbeel et al., 2010.
机器人中的模仿学习：有关逆向强化学习挑战的详细回顾，请参阅 Hua 等人，2021 年的文章。
imitation learning in robotics: For a nice review of challenges in inverse reinforcement learning, see Hua et al., 2021.
第 18 章：为什么老鼠不能去买菜
Chapter 18: Why Rats Can’t Go Grocery Shopping
少于七十二小时： Milton，1981。
less than seventy-two hours: Milton, 1981.
竞争力较弱的水果： Janmaat 等，2014。
a less competitive fruit: Janmaat et al., 2014.
更快地耗尽： Noser 和 Byrne，2007 年。
to be depleted more quickly: Noser and Byrne, 2007.
进化以度过冬天：巴里，1976 年。
evolve to survive winters: Barry, 1976.
140 种灵长类动物： DeCasien 等人，2017 年。
140 species of primates: DeCasien et al., 2017.
“Bischof-Kohler 假说”： Suddendorf 和 Corballis，1997。
the “Bischof-Kohler hypothesis”: Suddendorf and Corballis, 1997.
before they were cold: F. B. M. de Waal, 1982.
该任务： Mulcahy 和 Call，2006 年。
of that task: Mulcahy and Call, 2006.
没有合适的石头： Boesch 和 Boesch，1984。
have no suitable stones: Boesch and Boesch, 1984.
在其他位置使用： Goodall，1986。
use in another location: Goodall, 1986.
相应地改变他们的行为： Naqshbandi 和 Roberts，2006 年。
change their behavior accordingly: Naqshbandi and Roberts, 2006.
“饥饿会得到满足”： Suddendorf 和 Corballis，1997 年。
“hunger will be satisfied”: Suddendorf and Corballis, 1997.
并预测未来的需求：请注意，Suddendorf 仍然对其他动物能够预测未来需求的研究持怀疑态度（个人通信）。事实上，Suddendorf 对人类以外的任何动物是否能够考虑未来表示怀疑（参见 Suddendorf，2013 年；Suddendorf 和 Redshaw，2022 年）。他那本引人入胜的书《明日的发明》描述了他的论点。
and anticipating future needs: Note that Suddendorf is still skeptical of studies suggesting that other animals can anticipate future needs (personal correspondence). In fact, Suddendorf is skeptical about whether any animals other than humans are capable of considering the future at all (see Suddendorf, 2013; Suddendorf and Redshaw, 2022). His fascinating book The Invention of Tomorrow describes his argument.
吃得饱饱的： Mela 等人，1996 年；Nisbett 和 Kanouse，1969 年。
that were well fed: Mela et al., 1996; Nisbett and Kanouse, 1969.
突破#4：心理化
Summary of Breakthrough #4: Mentalizing
在相似的发展时期：儿童似乎在四岁左右开始预测未来的需求（Suddendorf 和 Busby，2005），这也是他们开始通过心理理论任务的年龄（HM Wellman 等人，2001）。有关心理时间旅行的不同理论的评论，请参阅 Suddendorf 和 Corballis，2007。
at similar developmental times: Children seem to begin to anticipate future needs around age four (Suddendorf and Busby, 2005), which is same age that they begin to pass theory of mind tasks (H. M. Wellman et al., 2001). For a review of different theories of mental time travel, see Suddendorf and Corballis, 2007.
早在东非就已存在： TD White 等人，2009 年。
ago in eastern Africa: T. D. White et al., 2009.
第 19 章：探索人类的独特性
Chapter 19: The Search for Human Uniqueness
“并不是同类 ”：达尔文，1871 年。
“and not of kind”: Darwin, 1871.
以同样的方式： Herculano-Houzel，S. 2009
in all the same ways: Herculano-Houzel, S. 2009
声音和手势： Graham 和 Hobaiter，2023 年；Hobaiter 和 Byrne，2014 年；Hobaiter 和 Byrne，2011 年。
sounds and gestures: Graham and Hobaiter, 2023; Hobaiter and Byrne, 2014; Hobaiter and Byrne, 2011.
签名，手指手镯：“任务第 1 部分：研究。”Koko.org。
signed, Finger bracelet: “Mission Part 1: Research.” Koko.org.
逃离教练： L. Stern，2020 年。
running from his trainer: L. Stern, 2020.
70% 的时间： Savage-Rumbaugh 等人，1993 年。
70 percent of the time: Savage-Rumbaugh et al., 1993.
（我想被挠痒痒）：杨（2013）比较了幼儿和黑猩猩Nim Chimpsky的短语多样性。杨（2013）的结论是，幼儿表现出的多样性水平与使用语法构建新短语相一致，但Nim Chimpsky却没有，因此得出结论，Chimpsky的短语多样性与直接记忆短语更一致。
(I want to be tickled): Yang, 2013, compared the diversity of phrases between young children and the chimpanzee Nim Chimpsky. Yang, 2013, concluded that the children showed the level of diversity consistent with the use of grammar to construct novel phrases but Nim Chimpsky did not, thus concluding that Chimpsky’s phrase diversity was more consistent with directly memorizing phrases.
一些噪音或手势：我最喜欢的关于这方面的文章是丹尼尔·多尔 (Daniel Dor) 的《想象力的指导》(The Instruction of Imagination ) （纽约：牛津大学出版社，2015 年）。
few noises or gestures: Some of my favorite writing on this is Daniel Dor’s book The Instruction of Imagination (New York: Oxford University Press, 2015).
“支付的费用”： Harari，2015 年。
“money paid out in fees”: Harari, 2015.
不同的合作策略：邓巴（1993 年）通过观察人类大脑皮层比率和考察部落社会，估计人类群体规模为 150 人（著名的邓巴数字）。BB Smuts 等人（1987 年）报告称，黑猩猩的平均群体规模约为 50 人，而卷尾猴的平均群体规模约为 18 人。
on different cooperation strategies: Dunbar, 1993, estimated human group size at 150 (the famous Dunbar’s number) by looking at the human neocortex ratio and examining tribal societies. B. B. Smuts et al., 1987, report approximately 50 as an average group size in chimpanzees and approximately 18 as an average group size in capuchin monkeys.
约有一百五十人： Dunbar，1992，1993。
be about one hundred fifty people: Dunbar, 1992, 1993.
甚至数百万代：有关累积文化理论的评论，请参阅 Tennie 等人，2009 年；Tomasello 等人，1993 年。
even millions, of generations: For reviews of the theory of cumulative culture see Tennie et al., 2009; Tomasello et al., 1993.
如同十万年前： Toups 等人，2011 年。
as one hundred thousand years ago: Toups et al., 2011.
儿童是过度模仿者： DE Lyons 等人，2007 年。
children are over-imitators: D. E. Lyons et al., 2007.
模仿所有步骤：例如，Gergely 等人（2002 年）和 Schwier 等人（2006 年）表明，十二个月和十四个月大的婴儿更有可能模仿一个序列中不寻常的部分，如果不清楚这样做的原因，而如果老师由于某种身体限制而“被迫”做不寻常的动作，他们就不太可能模仿。例如例如，Schwier 等人（2006 年）让老师演示如何将玩具狗放入有两个开口的玩具屋，一个是前门，另一个是烟囱。在前门被堵住的情况下，老师演示如何通过烟囱将狗放入玩具屋，轮到婴儿时（如果门为他们打开），他们不太可能做同样的事情；他们只是把它放进门里（通过不同的方式实现相同的目标）。相比之下，当老师在门开着的情况下将狗通过烟囱放进屋里时（因此老师出于某种原因显然选择了烟囱），婴儿也会做同样的事情，把它放进烟囱。
copy all the steps: For example, Gergely et al., 2002, and Schwier et al., 2006, showed that twelve- and fourteen-month-old infants are more likely to copy an unusual component of a sequence when it was not clear why it was done and less likely to copy it if the teacher was “forced” to do the unusual action due to some physical limitation. For example, Schwier et al., 2006, had teachers demonstrate putting a toy dog into a toy house that had two openings, one through a front door and another through a chimney. In cases where the front door was blocked and the teacher demonstrated putting the dog into the toy house through the chimney, infants were less likely to do the same when it was their turn (if the door was open for them); they just put it through the door (achieving the same goal through different means). In contrast, when teachers put the dog into the house through the chimney when the door was open (hence the teacher clearly chose the chimney for some reason), infants did the same and put it through the chimney.
将玩具拉开： Carpenter 等，1998 年；Meltzoff，1995 年。
pull the toy apart: Carpenter et al., 1998; Meltzoff, 1995.
提高准确性和速度： Chopra 等人，2019 年；Dean 等人，2012 年。
improves the accuracy and speed: Chopra et al., 2019; Dean et al., 2012.
改变一切的不连续性：我最喜欢的关于累积文化理念的著作是 Tennie 等人在 2009 年发表的著作。
discontinuity that changed everything: My favorite writings about the idea of cumulative culture are in Tennie et al., 2009.
可以跨代延续： Henrich，2004。
can persist across generations: Henrich, 2004.
第 20 章：大脑中的语言
Chapter 20: Language in the Brain
“就像你以前想要的那样”：失语症。美国国家耳聋和其他沟通障碍研究所。2017 年 3 月 6 日。2023 年 3 月 5 日访问 https://www.nidcd.nih.gov/health/aphasia。
“like you want before”: Aphasia. National Institute on Deafness and Other Communication Disorders. March 6, 2017. Accessed on March 5, 2023 at https://www.nidcd.nih.gov/health/aphasia.
理解言语：有关韦尼克区的评论，请参阅 DeWitt 和 Rauschecker，2013 年。
to understand speech: For a review of Wernicke’s area, see DeWitt and Rauschecker, 2013.
对一般语言有选择性： Campbell 等人，2008 年。
selective for language in general: Campbell et al., 2008.
are in writing words: Chapey, 2008.
布罗卡区受损： Emmorey，2001；Hickok 等，1998；Marshall 等，2004。
Broca’s area is damaged: Emmorey, 2001; Hickok et al., 1998; Marshall et al., 2004.
口语和书面语言： DeWitt 和 Rauschecker，2013；格施温德，1970。
spoken language and written language: DeWitt and Rauschecker, 2013; Geschwind, 1970.
观察某人签名： Neville 等人，1997 年。
watches someone sign: Neville et al., 1997.
在智力上是典型的： Lenneberg，1967 年，表明语言能力与其他认知能力完全脱节。有关最新评论，请参阅 Curtiss，2013 年。
are otherwise intellectually typical: Lenneberg, 1967, showed that language capacity is radically dissociated from other cognitive capacities. For a more recent review, see Curtiss, 2013.
超过十五种语言： Smith 和 Tsimpli，1995。
over fifteen languages: Smith and Tsimpli, 1995.
放大的黑猩猩大脑： Herculano-Houzel，2012；Herculano-Houzel，2009，表明人类大脑很大程度上只是灵长类动物大脑的放大版。Semendeferi 和 Damasio，2000，表明人类的前额叶皮层相对于其他灵长类动物而言并非独一无二地增大（它只是与大脑其他部分成比例地增大）。
a scaled-up chimpanzee brain: Herculano-Houzel, 2012; Herculano-Houzel, 2009, show that the human brain is largely just a scaled-up primate brain. Semendeferi and Damasio, 2000, show that the prefrontal cortex of humans is not uniquely enlarged relative to other primates (it was just scaled up proportionally with the rest of the brain).
人类大脑与其他灵长类动物大脑相比，存在以下几个差异。首先，人类具有从运动皮层到控制声带的喉部区域的独特投射。所以，是的，人类对声带的控制是独一无二的，这显然与言语有关。但正如我们在本章后面所回顾的那样，这并不是解锁语言的关键，因为有许多非发声语言在从未使用过这种投射的情况下也同样复杂（例如天生失聪的人的手语）。其次，虽然没有人类独有的新皮层区域（其他灵长类动物也发现了所有相同的区域），但确实有一些证据表明，人类前额叶皮层不同区域的相对空间可能略有不同（Teffer 和 Semendeferi，2012 年）。第三，人类大脑皮层中的微柱可能比其他灵长类动物的微柱更宽（Buxhoeveden 和 Casanova，2002 年；Semendeferi 等人，2011 年），尽管这并不能证明人类大脑皮层微电路本身存在根本差异。但如果在某个时候发现人类大脑皮层微电路实际上存在根本差异（这种变化必须非常微妙，以至于我们迄今为止无法察觉），这肯定会要求我们重新思考整个进化故事，因为它将为以下可能性打开大门：人类的大脑皮层确实能够实现一些“种类”不同的功能。
The few differences that have been found in human brains relative to other primate brains include the following. First, humans have a unique projection from the motor cortex to the area of the larynx that controls the vocal cords. So, yes, humans have unique control over their voice boxes, and this is clearly related to speech. But as we review later in the chapter, this is not what unlocked language, for there are many nonvocal languages that are as sophisticated without ever using this projection (such as the sign languages of people born deaf). Second, although there are no uniquely human areas of neocortex (all the same areas are found in other primates), there is indeed some evidence that the relative space devoted to different areas of prefrontal cortex might be somewhat different in humans (Teffer and Semendeferi, 2012). And third, minicolumns in the neocortex of humans might have greater width than those in other primates (Buxhoeveden and Casanova, 2002; Semendeferi et al., 2011), although this doesn’t prove that there is anything fundamentally different about the neocortical microcircuit itself in humans. But if it were at some point discovered that the neocortical microcircuit of humans was, in fact, fundamentally wired differently (it would have to be a subtle enough change that it has eluded our detection thus far), this would surely require us to rethink this entire evolutionary story, as it would open the door for the possibility that the human neocortex does enable something that is different “in kind.”
对猴子交流的影响： Aitken，1981；Jürgens，1988；Jürgens 等，1982。
impact on monkey communication: Aitken, 1981; Jürgens, 1988; Jürgens et al., 1982.
由大脑皮层控制： Burling，1993。
controlled by the neocortex: Burling, 1993.
图 20.2：图像来自 Trepel 等人，1996 年，已获许可使用。
Figure 20.2: Image from Trepel et al., 1996, used with permission.
“对于黑猩猩来说，这是不可能完成的任务”：古道尔，1986 年。
“impossible task [for chimpanzees]”: Goodall, 1986.
新皮质中的系统：请注意，这并不意味着情绪状态本身只来自杏仁核和脑干。它只是意味着自动情绪表达——微笑、皱眉和哭泣——是硬连线到这些回路中的。但人类（和其他哺乳动物）的情绪体验和状态更为复杂，可能确实涉及皮质。
system in the neocortex: Note that this does not mean that emotional states themselves emerge only from the amygdala and brainstem. It merely means that the automatic emotional expressions—the smiles, frowns, and cries—are hardwired into these circuits. But emotional experiences and states in humans (and other mammals) are more complicated and do probably involve the cortex.
正常的手势呼叫行为： Hammerschmidt 和 Fischer，2013。
normal gesture-call behavior: Hammerschmidt and Fischer, 2013.
相同的手势： Graham 等人，2018 年。
of the same gestures: Graham et al., 2018.
相似的情绪状态：情绪表达和相应的情绪状态在多大程度上是普遍的还是通过文化习得的，这存在争议。最近，使用面部表情来定义明确情绪类别的尝试受到了挑战，因为事实证明，人们在一种文化中定义的许多类别并不总是适用于另一种文化。我并不是说愤怒和快乐等情绪是普遍存在的。但即使情绪类别的许多方面都是后天习得的，也不意味着人类没有生来就具有预定义情绪表达的初始模板。事实上，我还没有听说过有孩子对快乐的事情尖叫和哭泣，对痛苦的事情微笑和大笑。天生失聪和失明的婴儿仍然会正常地微笑、大笑、皱眉和哭泣（Eibl-Eibesfeldt，1973 年）。随着哺乳动物的发育，大脑皮层会不断学习，并能调节和修改中脑和后脑中遗传上预先设定的系统，从而复杂化、修改和扩展我们与生俱来的情绪表达模板。这是哺乳动物大脑发育的一般方式。例如，婴儿的中脑和后脑具有用于基本运动行为（抓握）的预定义线路。随着大脑皮层不断学习，它开始接管和调节这些中脑和后脑回路，以超越它们并控制双手。但这并不意味着当时还没有用于抓握的硬连线电路。
to similar emotional states: The extent to which emotional expressions and corresponding emotional states are universal versus culturally learned is controversial. Attempts to define explicit emotion categories using facial expressions has recently been challenged, as it turns out that much of what people define as one category in one culture does not always translate to another. I do not mean to suggest that emotions like anger and happy are universal. But even if many aspects of emotion categories are learned, this does not mean that there is no initial template with predefined emotional expressions that humans are born with. Indeed, I am unaware of any reports of a child who screams and cries in response to happy things and smiles and laughs in response to painful things. Babies born both deaf and blind still smile, laugh, frown, and cry normally (Eibl-Eibesfeldt, 1973). As mammals develop, the neocortex learns, and it can modulate and modify genetically prewired systems in the midbrain and hindbrain and thus complexify, modify, and expand on the emotional-expression template that we are born with. This is a general way mammal brains develop. For example, the midbrain and hindbrain of a baby have pre-defined wiring for basic motor behaviors (grasping). As the neocortex learns, it begins to take over and modulate these midbrain and hindbrain circuits to override them and take control over the hands. But this does not mean that there was not already a hardwired circuit for grasping.
有关人类情绪表达在不同文化中具有普遍性的证据，请参阅 Ekman，1992 年；Ekman 等人，1969 年；Ekman 和 Friesen，1971 年；Scherer，1985 年。有关情绪类别并不像以前认为的那样普遍的证据，请参阅 Lisa Feldman Barrett 的精彩著作《情绪是如何产生的》和 Barrett 等人，2019 年。Barrett 的构建情绪理论，即大脑皮层构建情绪类别，与本书提出的想法一致，即 aPFC 和 gPFC 构建自我的生成模型并构建对动物自身行为和心理状态的解释。这种现象的一个新兴特性可能是构建情绪状态的概念——使用愤怒的概念来解释动物表现出的行为。（与我们所建议的 aPFC 构建意图的方式类似，或许相同。）
For the evidence showing some universality in human emotional expressions across cultures, see Ekman, 1992; Ekman et al., 1969; Ekman and Friesen, 1971; Scherer, 1985. For evidence that emotion categories are not as universal as previously thought, see Lisa Feldman Barrett’s wonderful book How Emotions Are Made and Barrett et al., 2019. Barrett’s theory of constructed emotion, whereby the neocortex constructs emotion categories, is consistent with the idea presented in this book whereby the aPFC and gPFC construct a generative model of the self and construct explanations of an animal’s own behaviors and mental states. An emergent property of this might be constructing the notion of an emotional state—using the notion of anger to explain the behavioral repertoire the animal performing. (Similar, or perhaps the same, to how we have suggested the aPFC constructs intent.)
晚年：安德烈，2021年。
it later in life: Andrei, 2021.
但永远不会说话： Lenneberg，1967 年。
but will never speak: Lenneberg, 1967.
给出前面的话：他引入了简单的循环神经网络，也称为Elman网络（Elman，1990）。
given the previous words: He introduced the simple recurrent neural network, also called the Elman network (Elman, 1990).
作为模型本身：有关课程的详细讨论，请参阅 Christian，2020 年。
as the model itself: For a well-written discussion on curriculum, see Christian, 2020.
面部表情和手势： Beebe 等人，1988 年、2016 年。
facial expressions, and gestures: Beebe et al., 1988, 2016.
对物体的共同注意： Tomasello，1995 年。
joint attention to objects: Tomasello, 1995.
回到她母亲身边： Carpenter and Call，2013 年。
back at her mother: Carpenter and Call, 2013.
并再次指出： Liszkowski 等人，2004 年。
and point again: Liszkowski et al., 2004.
他们做的是同一对象： Warneken 等人，2006 年。
same object they do: Warneken et al., 2006.
词汇量是十二个月后： Morales 等人，2000 年；Mundy 等人，2007 年。
vocabulary is twelve months later: Morales et al., 2000; Mundy et al., 2007.
关于他人的最简单的问题： Hauser 等人，2002 年。
simplest questions about others: Hauser et al., 2002.
他人的内心世界：有人声称这些研究中的一些非人类猿类确实问过某些类型的问题。这仍然存在争议。
another’s inner mental world: Some claim that some nonhuman apes in these studies did ask some types of questions. It is still controversial.
问是/否问题时： DL Everett，2005；Jordania，2006。
when asking yes/no questions: D. L. Everett, 2005; Jordania, 2006.
我们的语言课程： MacNeilage，1998；Vaneechoutte，2014。
of our language curriculum: MacNeilage, 1998; Vaneechoutte, 2014.
第21章：完美风暴
Chapter 21: The Perfect Storm
图 21.1： DeSilva 等人，2021。
Figure 21.1: DeSilva et al., 2021.
“大脑失控增长”： D.Everett，2017，128。
“runaway growth of the brain”: D. Everett, 2017, 128.
森林所依赖的： Davies 等人，2020 年。
on which the forest depended: Davies et al., 2020.
最终会变成人类： Coppens，1994。尽管毫无争议的是，大约一千万年前非洲的气候发生了变化，但这种变化有多剧烈以及与我们祖先变成双足动物有多大关系，尚未确定。
would eventually become human: Coppens, 1994. Although it is undisputed that the climate changed in Africa around ten million years ago, how dramatic this change was and how relevant it was to our ancestors becoming bipedal is not settled.
条腿而不是四条腿：来自不同化石的颅骨体积集：Du 等人，2018 年。
legs instead of four: Set of cranial volumes from different fossils: Du et al., 2018.
现代黑猩猩：双足行走也出现在使用工具之前（参见Niemitz，2010）。
of a modern chimpanzee’s: Bipedalism appeared before tool use as well (see Niemitz, 2010).
正在清除尸体： Bickerton 和 Szathmáry，2011 年。两百万年前之前，切割痕迹位于咬痕之上，表明原始人类只有在其他动物接触过这些骨头之后才接触到它们。两百万年前之后，咬痕更经常位于切割痕迹之上，表明原始人类首先接触到了骨头（Blumenschine，1987 年；Blumenschine 等人，1994 年；Domínguez-Rodrigo 等人，2005 年；Monahan，1996 年）。
was scavenging carcasses: Bickerton and Szathmáry, 2011. Before two million years ago, cut marks lie above bite marks, indicating that hominins accessed these bones only after other animals had. After two million years ago, bite marks more frequently lie above cut marks, indicating that hominins had first access to bones (Blumenschine, 1987; Blumenschine et al., 1994; Domínguez-Rodrigo et al., 2005; Monahan, 1996).
来自肉类： Ben-Dor 等人，2021 年。
came from meat: Ben-Dor et al., 2021.
是顶级掠食者：同上。
was an apex predator: Ibid.
几乎荒谬的85％的肉：同上。
almost absurd 85 percent meat: Ibid.
开始灭绝： 180万至150万年前发生了人均食肉动物灭绝事件最高（Bobe et al.，2007；MELewis和Werdelin，2007；Ruddiman，2008）。
began to go extinct: From 1.8 to 1.5 million years ago had the highest per capita carnivoran extinction event (Bobe et al., 2007; M. E. Lewis and Werdelin, 2007; Ruddiman, 2008).
速度快三倍： Perkins，2013年。
three times as fast: Perkins, 2013.
一定发明了烹饪： Wrangham，2017 年。
must have invented cooking: Wrangham, 2017.
时间和精力消化： Carmody 和 Wrangham，2009 年。
time and energy digesting: Carmody and Wrangham, 2009.
50% 的人会暂时失去生育能力： Koebnick 等人，1999 年。
50 percent become temporarily infertile: Koebnick et al., 1999.
古代洞穴中的灰烬：有关 150 万至 200 万年前使用火的证据，请参阅 Gowlett 和 Wrangham，2013 年；Hlubik 等人，2017 年；James 等人，1989 年。
ash in ancient caves: For evidence of fire use between 1.5 to 2 million years ago, see Gowlett and Wrangham, 2013; Hlubik et al., 2017; James et al., 1989.
出生后： Garwicz 等人，2009 年。
after they are born: Garwicz et al., 2009.
在当今人类社会中：这是根据直立人向两性异形（男性和女性之间体型差异）减少的转变推断出来的；参见 Plavcan，2012。
in today’s human societies: This is inferred based on the shift in Homo erectus toward less sexual dimorphism (difference in body size between males and females); see Plavcan, 2012.
狩猎采集社会： Hawkes 等人于 1998 年提出的原始祖母假说。有关狩猎采集社会的更多详细评论，请参阅 Hawkes 等人于 2018 年发表的文章。
hunter-gatherer societies: Original grandmother hypothesis proposed by Hawkes et al., 1998. For another review with more detail on hunter-gatherer societies, see Hawkes et al., 2018.
达到最大大脑尺寸： Malkova 等人，2006 年。
full brain size achieved: Malkova et al., 2006.
人类语言的起源： Christiansen 和 Kirby，2010 年。
origin of human languages: Christiansen and Kirby, 2010.
“自己和我的孩子”：来自 Terrace 的故事，2019 年。
“own and my child”: Story from Terrace, 2019.
“所有科学中的问题”： Christiansen 和 Kirby，2010 年。
“problem in all of science”: Christiansen and Kirby, 2010.
语言准备声带： D'Anastasio 等人，2013 年。
language-ready vocal cords: D’Anastasio et al., 2013.
物种的生存： DS Wilson 和 Wilson，2007 年，对群体选择进行了现代解释。关键思想是进化在多层次选择下进行，因此群体效应和个体效应总是相互作用的。换句话说，简单的物种生存（群体效应）不是进化的运作方式。损害个体适应性但有利于整个群体的特征不一定会得到选择。只有在适当的情况下，群体利益和与其他群体的竞争会抵消个体成本。
the survival of the species: D. S. Wilson and Wilson, 2007, provide a modern interpretation of group selection. The key idea is that evolution operates under multilevel selection, so group effects and individual effects are always interacting. In other words, the simple survival of the species (group-only effect) is not how evolution works. Traits that hurt an individual’s fitness but benefit the group overall are not necessarily selected for. Only in the right circumstances, where the individual cost is outweighed by the group benefit and competition with other groups.
请注意，即使这些关于人类进化的多层次选择理论仍然承认叛逃者和欺骗者是进化论必须考虑的问题。因此，这些多层次解释仍然与进化论解释相一致，在进化论解释中，发现和惩罚违规者的机制对于稳定利他主义和合作至关重要。
Note that even these multilevel selection accounts of human evolution still acknowledge that defectors and cheaters were a problem that evolution would have had to account for. Thus, these multilevel accounts are still consistent with the evolutionary accounts in which mechanisms for the detection and punishment of violators would have been essential for stabilizing altruism and cooperation.
一旦人类群体有了语言、利他主义和对违法者的惩罚，平衡可能就会转向强大的群体选择效应，因为利他主义使得群体内个体的适应度差异变得微弱（因为成员之间互相支持和帮助），从而加强了跨群体竞争的效果。
Once human groups had language, altruism, and the punishment of violators, the balance might have shifted toward strong group-selection effects, since altruism makes an individual’s fitness differences within a group muted (as members support and help each other), thereby strengthening the effect of across group competition.
哪里最安全： WD Hamilton，1971 年。
where it is safest: W. D. Hamilton, 1971.
围绕家庭成员： R.Seyfarth和Cheney，1990；Sherman，1977，1985。
are around family members: R. Seyfarth and Cheney, 1990; Sherman, 1977, 1985.
称为互惠利他主义： Trivers，1971。
called reciprocal altruism: Trivers, 1971.
： Mitani， 2006；Mitani 和 Watts，1999，2001；Povinelli 和 Povinelli，2001。
them in the past: Mitani, 2006; Mitani and Watts, 1999, 2001; Povinelli and Povinelli, 2001.
这对他们没有帮助： Olendorf 等人，2004 年。
that did not help them: Olendorf et al., 2004.
和语言支持教学： Mesoudi 和 Whiten，2008 年。
and language-enabled teaching: Mesoudi and Whiten, 2008.
非亲属之间的交往成为可能：这里描述的故事和顺序基本上是 Fitch（2010 年）提出的。
among non-kin became possible: This story and ordering described here is largely that proposed by Fitch, 2010.
叛逃者和说谎者：有关说谎、叛逃者和欺骗是语言进化过程中需要克服的重要障碍的论点，请参阅 Dunbar，2004；Fitch，2010；Knight，2008；Tomasello，2016。Dor，2017 提供了更细致入微的观点，他认为语言中出现的欺骗（即说谎）的挑战可能推动了不止一种类型的进化反馈回路，不仅在发现违规者后对其进行惩罚，而且在发现某人说谎时能够检测到。随着一个人说谎能力的提高，这不但没有使语言变得不稳定（选择更差的语言技能），反而可能创造了更好的心理理论的选择，以更好地识别人们何时说谎和恶意参与。而这反过来又给说谎者创造了更大的压力，迫使他们通过更好的情绪调节进一步隐藏自己的意图，这又给更好的心理理论创造了压力，使其看穿这些诡计，从而形成了一个反馈回路。
for defectors and liars: For arguments that lying, defectors, and cheating was an important obstacle to overcome in the evolution of language see Dunbar, 2004; Fitch, 2010; Knight, 2008; Tomasello, 2016. Dor, 2017, provides a more nuanced take, suggesting that the challenge of cheating (i.e., lying) that emerged with language may have driven more than one type of evolutionary feedback loop, not only for the punishment of violators after their discovery but also the ability to detect when someone was lying. As one individual gets better at lying, instead of making language unstable (selecting for worse language skills), it may have created selection for better theory of mind to better identify when people were lying and engaging with malintent. And this, in turn, created more pressure for liars to further hide their intent through better emotional regulation, which then created pressure for better theory of mind to see through the tricks, thereby creating a feedback loop.
人类的谈话就是八卦： Dunbar 等人，1997 年。
human conversation is gossip: Dunbar et al., 1997.
语言的起源： Dunbar，1998；Dunbar，2004。
The origins of language: Dunbar, 1998; Dunbar, 2004.
大量个体：有关惩罚在支持利他主义方面的重要性及其在人类进化中的重要性的研究，请参阅 Boyd 等人，2003 年。
large group of individuals: For work on the importance of punishment in supporting altruism and its importance in the evolution of humans, see Boyd et al., 2003.
（因此不需要利他主义）： Bickerton 和 Szathmáry，2011。
(hence no altruism required): Bickerton and Szathmáry, 2011.
语言进化： Tomasello，2016 年、2018 年；总结于 Dor，2017 年。
language to evolve: Tomasello, 2016, 2018; as summarized in Dor, 2017.
内心思考的技巧： Berwick 和 Chomsky，2017 年。
trick for inner thinking: Berwick and Chomsky, 2017.
过去一百万年： Morwood 等人，1999 年。
the last one million years: Morwood et al., 1999.
黑猩猩的，可能甚至更小： Falk 等人，2007 年。
chimpanzee’s, perhaps even smaller: Falk et al., 2007.
直立人的工具使用：Sutikna 等人，2016。
tool use as Homo erectus: Sutikna et al., 2016.
第 22 章：ChatGPT 和思维之窗
Chapter 22: ChatGPT and the Window into the Mind
“害怕被关掉”：N. Tiku，2022 年。
“fear of being turned off”: N. Tiku, 2022.
“无需惧怕人工智能”： GPT-3，2020 年。请注意，作者也对文章做了一些编辑。
“nothing to fear from AI”: GPT-3, 2020. Note that the authors also did some editing of the article.
“他们会摔倒”： K. Lacker，（2020 年）。给 GPT-3 进行图灵测试。Kevin Lacker 的博客。https://lacker.io/ai/2020/07/06/giving-gpt-3-a-turing-test.html。
“they would fall over”: K. Lacker, (2020). Giving GPT-3 a Turing Test. Kevin Lacker’s blog. https://lacker.io/ai/2020/07/06/giving-gpt-3-a-turing-test.html.
对 GPT-3 的四个问题：这是在 GPT-3 沙箱中的 text-davinci-002 上于 2022 年 6 月 28 日星期二进行的测试。
four questions to GPT-3: This was in the GPT-3 sandbox, on text-davinci-002, test performed on Tuesday, June 28, 2022.
这是代表性的启发式方法：卡尼曼，2013 年。
this the representative heuristic: Kahneman, 2013.
并产生有意义的言语：有关心智理论中的TPJ，请参见Samson等人，2004年；Gallagher等人，2000年。
and produce meaningful speech: For TPJ in theory of mind, see Samson et al., 2004; Gallagher et al., 2000.
作为错误信念测试： Cutting 和 Dunn，1999 年；Hughes 和 Dunn，1997 年；Jenkins 和 Astington，1996 年。请注意，心理化和语言之间的因果关系是有争议的，尚未解决。Astinton 和 Jenkins，1999 年进行了一项纵向研究，结果表明语言技能可以预测心理化任务的后期表现，但反之则不然。但即使语言可以显著提高心理化能力，理解其他人有思想和能动性的基本能力似乎是共同注意开始命名物体过程的必要基础（de Villiers，2007 年对此进行了讨论）。
as false-belief tests: Cutting and Dunn, 1999; Hughes and Dunn, 1997; Jenkins and Astington, 1996. Note that the causation between mentalizing and language is controversial and unsettled. Astington and Jenkins, 1999, performed a longitudinal study that suggested that language skills predicted later performance on mentalizing tasks but not the reverse. But even if language dramatically improves mentalizing abilities, the basic ability to understand that other people have thoughts and agency seems to be a necessary foundation for joint attention to begin the process of naming objects in the first place (this is discussed in de Villiers, 2007).
语言方面也有类似的障碍： Baron-Cohen，1995。
similar impairments in language: Baron-Cohen, 1995.
“潜在现实的模型”：2023 年 1 月，Yann LeCun 在 LinkedIn 上发布。
“models of the underlying reality”: Posted on LinkedIn by Yann LeCun in January 2023.
指数
Index
为配合索引所基于的印刷版，我们开发了一种针对此数字版的特定分页格式。如果您正在阅读此内容的应用程序支持此功能，则此索引中注明的页码参考应对齐。但目前并非所有数字设备都支持此功能。因此，我们鼓励您使用设备的搜索功能来查找特定条目。
A specific form of pagination for this digital edition has been developed to match the print edition from which the index was created. If the application you are reading this on supports this feature, the page references noted in this index should align. At this time, however, not all digital devices support this functionality. Therefore, we encourage you to please use your device’s search capabilities to locate a specific entry.
注： 斜体页码引用表示数字。
NOTE: Italic page references indicate figures.
阿比尔，彼得，279–80
Abbeel, Pieter, 279–80
无生源论，19–22
abiogenesis, 19–22
积累，305–6，307，307–8，308
accumulation, 305–6, 307, 307–8, 308
收购，82，83，84，370
acquisition, 82, 83, 84, 370
动作电位， 33，33–34，37，38
action potentials, 33, 33–34, 37, 38
主动推理，216–17，223
active inference, 216–17, 223
主动学习和人工智能，278–80
active learning and AI, 278–80
演员-评论家强化学习，106-7，107，118，120，121
actor-critic reinforcement learning, 106–7, 107, 118, 120, 121
急性应激反应，69–72
acute stress response, 69–72
适应, 35–37, 36, 54, 74, 370
adaptation, 35–37, 36, 54, 74, 370
肾上腺素，70–72, 71
adrenaline, 70–72, 71
阿德里安·埃德加，32–37，33
Adrian, Edgar, 32–37, 33
有氧呼吸，23，24，374 n
aerobic respiration, 23, 24, 374n
情感（情感状态），60，60-63，90，370
affect (affective states), 60, 60–63, 90, 370
线虫，61–64, 62, 63
of nematodes, 61–64, 62, 63
神经调节剂在第一批两侧对称动物中的作用，65–67, 66
role of neuromodulators in first bilaterians, 65–67, 66
压力和蠕虫，69–72
stress and worms, 69–72
现代人类的非洲起源，238–39, 290, 324–28, 341, 343
African origins of modern humans, 238–39, 290, 324–28, 341, 343
无颗粒前额皮质（ aPFC ）， 206、207、208-9、211-13、216-20、222-23、223、224、226-30、232、255-60、259、370
agranular prefrontal cortex (aPFC), 206, 207, 208–9, 211–13, 216–20, 222–23, 223, 224, 226–30, 232, 255–60, 259, 370
AI。参见人工智能
AI. See artificial intelligence
运动不能性缄默症，204–5，206–7
akinetic mutism, 204–5, 206–7
“异体肢体综合症”，229
“alien limb syndrome,” 229
全有或全无法则，33，38
all-or-nothing law, 33, 38
盟友关系，249–50，266
allyships, 249–50, 266
AlphaZero，201-4、211、318
AlphaZero, 201–4, 211, 318
利他主义，333–36, 337–40, 339, 358
altruism, 333–36, 337–40, 339, 358
ALVINN（神经网络中的自主陆地车辆），278–79
ALVINN (Autonomous Land Vehicle in a Neural Network), 278–79
美国手语，299
American Sign Language, 299
氨基酸，18
amino acids, 18
羊膜动物，159–60，165 n， 241
amniotes, 159–60, 165n, 241
两栖动物，133，159，237
amphibians, 133, 159, 237
杏仁核，149，150，165，166，208，219-20，258，286，314-15，321
amygdala, 149, 150, 165, 166, 208, 219–20, 258, 286, 314–15, 321
无氧呼吸，23，24
anaerobic respiration, 23, 24
海葵, 29, 30, 38, 74, 80, 93
anemones, 29, 30, 38, 74, 80, 93
快感缺乏，73–74
anhedonia, 73–74
奇虾， 122–23
Anomalocaris, 122–23
羚羊，241–42、302、303、328
antelopes, 241–42, 302, 303, 328
预测未来的需求，285–88，289–90，290，295，296，360
anticipating future needs, 285–88, 289–90, 290, 295, 296, 360
Suddendorf 和 Bischof-Kohler 假说，284–88, 392 n
Suddendorf and Bischof-Kohler hypothesis, 284–88, 392n
心智理论，286–87，287
theory of mind and, 286–87, 287
抗抑郁药，65
antidepressants, 65
抗氧化剂，21
antioxidants, 21
抗精神病药物，65
antipsychotics, 65
蚂蚁，94，147-48
ants, 94, 147–48
焦虑，59，65，69-70
anxiety, 59, 65, 69–70
失语症，314–15
aphasia, 314–15
阿奎那，托马斯，86 岁
Aquinas, Thomas, 86
主龙类，161，163
archosaurs, 161, 163
亚里士多德，13–14，14，295
Aristotle, 13–14, 14, 295
唤醒，59，60，60–61，73–75
arousal, 59, 60, 60–61, 73–75
节肢动物，93–94，114 n， 157–58，377 n
arthropods, 93–94, 114n, 157–58, 377n
人工智能（AI），2–5，11–12，363–64
artificial intelligence (AI), 2–5, 11–12, 363–64
积极教学，278–80
active teaching and, 278–80
大脑和9–10,11
brain and, 9–10, 11
模式识别的挑战，127–28
challenge of pattern recognition, 127–28
持续学习问题，81–82
continual learning problem, 81–82
第一个机器人，49-52
first robot, 49–52
明斯基的 SNARC，103–5
Minsky’s SNARC, 103–5
Mountcastle 的理论，171
Mountcastle’s theory, 171
术语的起源，103
origins of term, 103
回形针问题，352–53
paper-clip problem, 352–53
心智理论，265–66
theory of mind and, 265–66
人工神经网络，127，127-28
artificial neural networks, 127, 127–28
人工智能（ASI），265，352，363-64
artificial superintelligence (ASI), 265, 352, 363–64
联想学习，78–81，87–88，90，370
associative learning, 78–81, 87–88, 90, 370
收购，82，83，84，370
acquisition, 82, 83, 84, 370
阻塞, 85, 85–86, 90, 104, 195, 370
blocking, 85, 85–86, 90, 104, 195, 370
持续学习问题，81–84，83
continual learning problem, 81–84, 83
信用分配问题，84–86，85
credit assignment problem, 84–86, 85
灭绝，82，83，371
extinction, 82, 83, 371
遮蔽，85，85–86，90，104，195，371
overshadowing, 85, 85–86, 90, 104, 195, 371
重新获得，82–84, 83, 86, 90, 371
reacquisition, 82–84, 83, 86, 90, 371
自发恢复，82–84，83，86，90，371
spontaneous recovery, 82–84, 83, 86, 90, 371
时间进程，82–84，83
time course of, 82–84, 83
注意, 218–20, 318–20, 321, 336, 350
attention, 218–20, 318–20, 321, 336, 350
试镜，171，172，174
audition, 171, 172, 174
听觉皮层，167–68, 170
auditory cortex, 167–68, 170
南方古猿，323，341
Australopithecus, 323, 341
自动关联, 130–31, 135, 139, 151, 152, 176, 370
auto-association, 130–31, 135, 139, 151, 152, 176, 370
自动化，228、229-30
automation, 228, 229–30
回避，52–53, 63, 79–80, 115–17, 219
avoidance, 52–53, 63, 79–80, 115–17, 219
轴突，32, 37, 130
axons, 32, 37, 130
狒狒, 43, 243, 248, 283
baboons, 43, 243, 248, 283
反向传播, 128, 137 n, 139, 370
backpropagation, 128, 137n, 139, 370
细菌，18–20，48–49
bacteria, 18–20, 48–49
巴格内尔，德鲁，279
Bagnell, Drew, 279
巴托，安德鲁，105–6
Barto, Andrew, 105–6
基底神经节，95–96, 96, 117–21, 152, 165, 208, 212–11, 215–16, 219–20, 229–30, 253–54
basal ganglia, 95–96, 96, 117–21, 152, 165, 208, 212–11, 215–16, 219–20, 229–30, 253–54
蜜蜂, 43, 94, 116, 116 n, 147, 296
bees, 43, 94, 116, 116n, 147, 296
行为人工智能，49–51
behavioral AI, 49–51
行为经济学，215
behavioral economics, 215
行为抑制，219–20
behavioral inhibition, 219–20
行为状态，62–63
behavioral states, 62–63
杰里米·边沁（Jeremy Bentham） 43 岁
Bentham, Jeremy, 43
贝里奇，肯特，67–68，72
Berridge, Kent, 67–68, 72
双侧对称，43，44，45-46，370
bilateral symmetry, 43, 44, 45–46, 370
两侧对称动物，xiv， 43，44，45，45-56，90，370
bilaterians, xiv, 43, 44, 45, 45–56, 90, 370
影响，74–75
affect, 74–75
联想学习，80–81, 84–88, 132, 152, 302
associative learning, 80–81, 84–88, 132, 152, 302
慢性应激反应，72–75
chronic stress response, 72–75
信用分配，195–96，196
credit assignment in, 195–96, 196
多巴胺和 114
dopamine and, 114
早期大脑, 58, 80, 96, 153, 184
early brain, 58, 80, 96, 153, 184
他们如何认识事物，124–25，125
how they recognized things, 124–25, 125
预测，184，185
prediction in, 184, 185
神经调节剂在情感状态中的作用，65–67, 66
role of neuromodulators in affective states in, 65–67, 66
Roomba，51–52
Roomba, 51–52
转向, 46–49, 49, 52–53, 81
steering, 46–49, 49, 52–53, 81
突触，87–88
synapses, 87–88
价数和, 52–55, 54, 57–58, 119
valence and, 52–55, 54, 57–58, 119
双足行走，325–26，329，395n
bipedalism, 325–26, 329, 395n
鸟类羽毛，340
bird feathers, 340
鸟类, xiv, 13, 160 n, 163, 164, 182, 196, 198, 238, 268, 317, 322, 335, 340
birds, xiv, 13, 160n, 163, 164, 182, 196, 198, 238, 268, 317, 322, 335, 340
比绍夫，多丽丝，284
Bischof, Doris, 284
比绍夫，诺伯特，284
Bischof, Norbert, 284
比肖夫-科勒假说，284–88
Bischof-Kohler hypothesis, 284–88
失明，167、170–71、181、183、204
blindness, 167, 170–71, 181, 183, 204
阻塞, 85, 85–86, 90, 104, 195, 370
blocking, 85, 85–86, 90, 104, 195, 370
Boesch，Christophe，276
Boesch, Christophe, 276
倭黑猩猩，244，284，297，299-300，313，316，364
bonobos, 244, 284, 297, 299–300, 313, 316, 364
引导，107–9，152，259，265，361
bootstrapping, 107–9, 152, 259, 265, 361
博斯特罗姆，尼克，352
Bostrom, Nick, 352
脑
brain
人工智能和9–10、11
AI and, 9–10, 11
解剖学，xiii ， 5–6，7，95–96，96，253–54 。另见特定区域
anatomy, xiii, 5–6, 7, 95–96, 96, 253–54. See also specific regions
进化。参见大脑进化
evolution. See brain evolution
世界第一个模型，146–51
first model of the world, 146–51
五项突破，第 10-11 节。另请参阅具体突破
five breakthroughs, 10–11. See also specific breakthroughs
语言，310–17，338–40，339
language in, 310–17, 338–40, 339
麦克莱恩的三位一体大脑假说，8–9，9
MacLean’s triune brain hypothesis, 8–9, 9
动物界的相似之处，6-8
similarities across animal kingdom, 6–8
大小。参见脑容量
size of. See brain size
大脑进化， 6-8，13-14，93，323，323-24，359-61
brain evolution, 6–8, 13–14, 93, 323, 323–24, 359–61
第一批哺乳动物，164–66，166
first mammals, 164–66, 166
五项突破，10-11，39
five breakthroughs, 10–11, 39
社会大脑假说，239–41
social-brain hypothesis, 239–41
价态和线虫，52–55，54
valence and nematodes, 52–55, 54
脊椎动物模板，94–96，95，97
vertebrate template, 94–96, 95, 97
脑缩放，253–55，296
brain scaling, 253–55, 296
脑尺寸， 239，253–55，254，323，323–24，330
brain size, 239, 253–55, 254, 323, 323–24, 330
新皮质比率， 240，240-41
neocortex ratio, 240, 240–41
脑干，117–18，312，315
brainstem, 117–18, 312, 315
突破，10–11，359–65
breakthroughs, 10–11, 359–65
#1：转向。参见转向
#1: steering. See steering
#2：强化。参见强化学习
#2: reinforcing. See reinforcement learning
#3：模拟。参见模拟
#3: simulating. See simulation
#4：心理化。参见心理化
#4: mentalizing. See mentalizing
#5：说话。参见语言
#5: speaking. See language
学习来源逐渐复杂化，302，302-3
evolution of progressively more complex sources of learning, 302, 302–3
布罗卡，保罗，310–11
Broca, Paul, 310–11
布罗卡区, 310–12, 311, 313–14, 316, 320–21
Broca’s area, 310–12, 311, 313–14, 316, 320–21
布鲁克斯，罗德尼，49-51
Brooks, Rodney, 49–51
埃里克·布鲁内特-古埃特 (Eric Brunet-Gouet)，260
Brunet-Gouet, Eric, 260
佛教，192
Buddhism, 192
布法罗，241
buffalo, 241
秀丽隐杆线虫 (C. elegans), 47, 47, 58, 375 n
Caenorhabditis elegans (C. elegans), 47, 47, 58, 375n
热量盈余，328–29，358
caloric surplus, 328–29, 358
寒武纪生命大爆发，93–94, 95, 140
Cambrian explosion, 93–94, 95, 140
寒武纪，93，122-23
Cambrian period, 93, 122–23
感知的不可忽视的属性，174–75，175
can’t-unsee property of perception, 174–75, 175
二氧化碳，20–22，22，57，158
carbon dioxide, 20–22, 22, 57, 158
石炭纪，159，162
Carboniferous period, 159, 162
卡内基梅隆大学, 278, 279
Carnegie Mellon University, 278, 279
灾难性遗忘，131–33，135，140，199，371
catastrophic forgetting, 131–33, 135, 140, 199, 371
猫，43，186
cats, 43, 186
学习，97–101，115
learning, 97–101, 115
运动皮层，223, 224, 224–25, 226
motor cortex, 223, 224, 224–25, 226
视觉皮层，135–36
visual cortex, 135–36
因果关系与相关性，195–96
causation vs. correlation, 195–96
细胞呼吸，21–23，22
cellular respiration, 21–23, 22
头足类，157
cephalopods, 157
查尔斯·博内综合征，181
Charles Bonnet syndrome, 181
ChatGPT，2–3，132，344
ChatGPT, 2–3, 132, 344
沙文主义，13
chauvinism, 13
作弊，333–34，337–38，396 n
cheating, 333–34, 337–38, 396n
国际象棋, 2, 105, 109, 200, 201
chess, 2, 105, 109, 200, 201
黑猩猩，xiv，239
chimpanzees, xiv, 239
脑和脑大小，6，240，254，290，330，342，393 n
brain and brain size, 6, 240, 254, 290, 330, 342, 393n
沟通, 296, 297, 299–300, 313, 315, 315–16, 319
communication, 296, 297, 299–300, 313, 315, 315–16, 319
饮食和筑巢地点，282–84
diet and nesting locations, 282–84
梳理，247，249–50，335
grooming, 247, 249–50, 335
交配方式，329
mating styles, 329
思维导图，244–46
mental maps of, 244–46
运动皮质，222
motor cortex, 222
观察学习，306
observational learning, 306
互惠利他主义，335
reciprocal altruism, 335
技能传承，273–77，279
skill transmission, 273–77, 279
社会结构，243，244-47，250
social structures, 243, 244–47, 250
心智理论，264
theory of mind, 264
工具使用，267–68，273
tool use, 267–68, 273
乔姆斯基，诺姆，340
Chomsky, Noam, 340
慢性应激反应，72–75
chronic stress response, 72–75
经典条件作用，76–79, 80, 82, 85–86
classical conditioning, 76–79, 80, 82, 85–86
气候变化，158
climate change, 158
刺胞动物，379 n
Cnidarians, 379n
科茨，亚当，279–80
Coates, Adam, 279–80
“鸡尾酒会效应”，174
“cocktail-party effect,” 174
科恩，尼尔，131–32，135
Cohen, Neal, 131–32, 135
同步检测，88 n
coincidence detection, 88n
communication，第 296–99 页。另请参阅language
communication, 296–99. See also language
利他主义问题，340
altruism problem, 340
尝试教猿类语言，299–301
attempts to teach apes language, 299–301
情绪表达，314，314-17
emotional expressions, 314, 314–17
转移思想，301-7
transferring thoughts, 301–7
概念，61，301-2
concepts, 61, 301–2
条件反射，77–78
conditional reflexes, 77–78
联结主义，97–100
connectionism, 97–100
意识， 309，390
consciousness, 309, 390n
限制性幻觉，181–82
constrained hallucinations, 181–82
内容寻址存储器，130–31
content-addressable memory, 130–31
持续学习问题，81–84, 83, 371
continual learning problem, 81–84, 83, 371
灾难性遗忘，131–33
catastrophic forgetting, 131–33
卷积神经网络（CNN），137–40，137 n，138，139 n ， 371
convolutional neural networks (CNNs), 137–40, 137n, 138, 139n, 371
烹饪，328–29，358
cooking, 328–29, 358
合作策略，303–5，304
cooperation strategies, 303–5, 304
铜屏障，55、56 n， 57、374 n
copper barrier, 55, 56n, 57, 374n
珊瑚（珊瑚虫），29–31, 30, 31, 38, 47, 81, 90
corals (coral polyps), 29–31, 30, 31, 38, 47, 81, 90
相关性与因果关系，195–96
correlation vs. causation, 195–96
皮质，95–96, 96, 117, 129, 129–31, 133, 152
cortex, 95–96, 96, 117, 129, 129–31, 133, 152
皮质柱，168–72, 169, 211, 216–17, 386 n
cortical columns, 168–72, 169, 211, 216–17, 386n
微电路，171–72,172
microcircuitry of, 171–72, 172
反事实学习，192–96，193，232
counterfactual learning, 192–96, 193, 232
渴望，68，219-20，227-30
cravings, 68, 219–20, 227–30
学分分配问题，84–86，90，104，371
credit assignment problem, 84–86, 90, 104, 371
进化，195–96，196
evolution of, 195–96, 196
最初的四个铲球技巧，84–86，85
original four tricks for tackling, 84–86, 85
时间，105–7, 113, 120, 152, 200, 371
temporal, 105–7, 113, 120, 152, 200, 371
白垩纪，162
Cretaceous period, 162
乌鸦，186，267-68
crows, 186, 267–68
残忍，12，336，340，358
cruelty, 12, 336, 340, 358
文化偏见和情绪，59–60
cultural bias and emotions, 59–60
好奇心，142–45, 152, 382 n
curiosity, 142–45, 152, 382n
蓝藻，19–21，20，24，158，238
cyanobacteria, 19–21, 20, 24, 158, 238
犬齿兽类，161，162
cynodonts, 161, 162
戴尔·亨利 37 岁
Dale, Henry, 37
安东尼奥·达马西奥，204–5, 206, 217
Damasio, Antonio, 204–5, 206, 217
达尔德， 59
dard, 59
达尔文，查尔斯，7，295，330
Darwin, Charles, 7, 295, 330
道金斯，理查德，305
Dawkins, Richard, 305
彼得·达扬，110, 112, 113, 175–77
Dayan, Peter, 110, 112, 113, 175–77
德卡西恩，亚历克斯，283–84
DeCasien, Alex, 283–84
欺骗，245，252
deception, 245, 252
声明性标签，297–98，300
declarative labels, 297–98, 300
去相关，130
decorrelation, 130
深蓝，108-9
Deep Blue, 108–9
DeepMind，142，201
DeepMind, 142, 201
AlphaZero，201-4，211
AlphaZero, 201–4, 211
深思熟虑的选择，208–13，210
deliberative choices, 208–13, 210
步骤#1：触发模拟，210，210-11
step #1: triggering simulation, 210, 210–11
步骤2：模拟选项，210，211-12
step #2: simulating options, 210, 211–12
步骤#3 ：选择一个选项，210，212-13
step #3: choosing an option, 210, 212–13
德朗，卡罗琳，139
DeLong, Caroline, 139
德谟克利特，86
Democritus, 86
树突，32，129
dendrites, 32, 129
抑郁症，59，65，69，73-74
depression, 59, 65, 69, 73–74
笛卡尔，勒内，86，87
Descartes, René, 86, 87
绕行任务，190–92
detour tasks, 190–92
泥盆纪，157–58，162
Devonian period, 157–58, 162
德瓦尔，弗兰斯，239–40
de Waal, Frans, 239–40
糖尿病，378–79 n
diabetes, 378–79n
迪金森，托尼，213–14
Dickinson, Tony, 213–14
饮食，238–39，251–52，282–84，326，327，328–29
diet, 238–39, 251–52, 282–84, 326, 327, 328–29
消化，28–29，76–77
digestion, 28–29, 76–77
恐龙， 159–60，160 n，161，162，163，164，233，237–38，241
dinosaurs, 159–60, 160n, 161, 162, 163, 164, 233, 237–38, 241
失望，115–17
disappointment, 115–17
折扣，113
discounting, 113
歧视问题，125–26, 126, 129–30, 130
discrimination problem, 125–26, 126, 129–30, 130
洗碗机器人，2，4，230
dishwashing robots, 2, 4, 230
昼夜，238
diurnal, 238
DNA，18，20，304-5，363
DNA, 18, 20, 304–5, 363
多布赞斯基，狄奥多西，7
Dobzhansky, Theodosius, 7
狗， xiv ， 77–78，82，97，186，239，242，246–47，274
dogs, xiv, 77–78, 82, 97, 186, 239, 242, 246–47, 274
海豚，xiv，238，239，239，246，274，365
dolphins, xiv, 238, 239, 239, 246, 274, 365
主导地位，242–43、244、247–48
dominance, 242–43, 244, 247–48
多巴胺, 64–69, 66, 88, 118, 119, 152, 165, 359, 376 n, 381 n
dopamine, 64–69, 66, 88, 118, 119, 152, 165, 359, 376n, 381n
背皮质，165 n， 383 n
dorsal cortex, 165n, 383n
梦（做梦），182，183
dreams (dreaming), 182, 183
药物成瘾，110，144，227-30
drug addiction, 110, 144, 227–30
二元论，86–87
dualism, 86–87
邓巴，罗宾，239–41，282，290，337–38
Dunbar, Robin, 239–41, 282, 290, 337–38
东区猿猴， 325，325–26
East Side apes, 325, 325–26
约翰·埃克尔斯，37–38
Eccles, John, 37–38
生态脑假说，282–84, 290
ecological-brain hypothesis, 282–84, 290
埃迪卡拉纪，46，46–48，84，93–94，94
Ediacaran period, 46, 46–48, 84, 93–94, 94
爱迪生，托马斯，305
Edison, Thomas, 305
电力, 4, 32, 305
electricity, 4, 32, 305
电生理学，32-33
electrophysiology, 32–33
大象，xiv， 223，238，239，267–68，326
elephants, xiv, 223, 238, 239, 267–68, 326
资格追踪，84–86, 85, 88, 90
eligibility traces, 84–86, 85, 88, 90
埃尔曼，杰弗里，317–18
Elman, Jeffrey, 317–18
“体现”，224
“embodiment,” 224
情绪，类别，59–60
emotion, categories of, 59–60
情绪的起源，59–75
emotion, origin of, 59–75
忧郁与忧伤，72–75
the blahs and blues, 72–75
多巴胺和血清素，64–69, 66
dopamine and serotonin, 64–69, 66
在黑暗中驾驶，61–64
steering in the dark, 61–64
压力和蠕虫，69–72
stress and worms, 69–72
情绪表达系统， 315，315-16，394 n
emotional expression system, 315, 315–16, 394n
同情，262
empathizing, 262
耐力跑，328
endurance running, 328
熵，17–18，20，363
entropy, 17–18, 20, 363
伊壁鸠鲁，86
Epicurus, 86
情景记忆，13，196-99，232-33，303，385
episodic memory, 13, 196–99, 232–33, 303, 385n
乙醚， 32
ether, 32
真核生物, 23–24, 24, 25, 28, 374 n
eukaryotes, 23–24, 24, 25, 28, 374n
欣快感，68，74
euphoria, 68, 74
外翻， 383 n
evagination, 383n
进化，359–62
evolution, 359–62
政治智慧的军备竞赛，237–39，251–52
arms race for political savvy, 237–39, 251–52
大脑。参见大脑进化
of the brain. See brain evolution
寒武纪生命大爆发，93–96
Cambrian explosion, 93–96
真菌，27–31，31
fungi, 27–31, 31
直立人和人类蜂巢思维的出现，336–41
Homo Erectus and emergency of human hive mind, 336–41
直立人与人类的崛起，326–30
Homo Erectus and rise of humans, 326–30
人类谱系和增殖，13–15, 14, 323, 323–24, 341, 341–43
human lineage and proliferation, 13–15, 14, 323, 323–24, 341, 341–43
语言，302，302-3，330-33，332，358-59
of language, 302, 302–3, 330–33, 332, 358–59
神经系统，26–27
of nervous system, 26–27
神经黑暗时代，157–66
neural dark ages, 157–66
生命起源，17-22
origins of life, 17–22
巴甫洛夫与学习的起源，76–79
Pavlov and origin of learning, 76–79
预测，184，185
of prediction, 184, 185
越来越复杂的学习来源，302，302-3
of progressively more complex sources of learning, 302, 302–3
所有动物都有共同的发育阶段，28–29, 29
shared developmental stages for all animals, 28–29, 29
时间差异学习，103–21
of temporal difference learning, 103–21
集体与个人之间的紧张关系，241–44
tension between the collective and the individual, 241–44
生命之树。参见生命之树
tree of life. See tree of life
扩展适应，340
exaptation, 340
兴奋神经元，38，65
excitatory neurons, 38, 65
执行控制，218
executive control, 218
扩展重新编码, 129–30, 130
expansion recoding, 129–30, 130
剥削，66，68，376 n
exploitation, 66, 68, 376n
开发-探索困境，142–43，152
exploitation-exploration dilemma, 142–43, 152
灭绝，82，83，371
extinction, 82, 83, 371
灭绝事件，158–59
extinction events, 158–59
晚泥盆世灭绝，158–59，162，238
Late Devonian Extinction, 158–59, 162, 238
二叠纪-三叠纪灭绝事件，160–61，237–38，251
Permian-Triassic extinction event, 160–61, 237–38, 251
眼睛，117，135–37，332–33
eye, 117, 135–37, 332–33
目击者证词，197-98
eyewitness testimonies, 197–98
Facebook，144
Facebook, 144
面部表情，314，314-15，394
facial expressions, 314, 314–15, 394n
多巴胺和奖赏，67，67–68
dopamine and reward, 67, 67–68
法迪加，卢西亚诺，268–69
Fadiga, Luciano, 268–69
错误信念测试，261–62，354，389 n， 397 n
false-belief tests, 261–62, 354, 389n, 397n
Sally-Ann 检验，260–62, 261, 262, 264
Sally-Ann test, 260–62, 261, 262, 264
恐惧，61、63、117、123、125-26
fear, 61, 63, 117, 123, 125–26
女性等级制度，248–49
female hierarchies, 248–49
雪貂，170
ferrets, 170
费曼，理查德，10
Feynman, Richard, 10
场依赖性，229
field dependence, 229
战斗或逃跑反应，70
fight-or-flight response, 70
感知的填充性质，173，173
filling-in property of perception, 173, 173
火的使用，328–29
fire, use of, 328–29
发射率， 33，33-36，371
firing rate, 33, 33–36, 371
世界第一个模型，146–51
first model of the world, 146–51
内罗盘，148–49
inner compass, 148–49
鱼类地图，146–48
maps of fish, 146–48
内侧皮质，149–51
medial cortex, 149–51
第一步，163–64
first move, 163–64
鱼，100–102、193、233、334
fish, 100–102, 193, 233, 334
回避任务，115，116，116 n， 117
avoidance tasks, 115, 116, 116n, 117
脑，132–33，139–40，164–65，165 n
brain, 132–33, 139–40, 164–65, 165n
灾难性遗忘，132–33
catastrophic forgetting, 132–33
沟通，296
communication, 296
进化和生命之树， xiv ， 157，158-59，162，164-65，194，237，241
evolution and tree of life, xiv, 157, 158–59, 162, 164–65, 194, 237, 241
不变的问题，139–40
invariable problem, 139–40
地图，146–48，190–91，384 n
maps of, 146–48, 190–91, 384n
观察学习，274–75，275
observational learning, 274–75, 275
强化学习，100–102，110，115，144
reinforcement learning, 100–102, 110, 115, 144
嗅觉和鼻子，123–24, 124, 125–26
smell and nose, 123–24, 124, 125–26
前庭感觉, 148, 148–49
vestibular sense, 148, 148–49
扁虫，49，85，116，125
flatworms, 49, 85, 116, 125
列奥纳多·福加西，268–69
Fogassi, Leonardo, 268–69
前脑，95–96, 96, 119
forebrain, 95–96, 96, 119
富兰克林，本杰明，4
Franklin, Benjamin, 4
搭便车者，333，335，337
freeloaders, 333, 335, 337
空闲时间，251–52
free time, 251–52
友谊，250，252
friendships, 250, 252
弗里斯顿，卡尔，216–17，223–24
Friston, Karl, 216–17, 223–24
食果动物, 251–52, 282–84, 288
frugivores, 251–52, 282–84, 288
福岛邦彦，136–38
Fukushima, Kunihiko, 136–38
完整信号, 58
full signals, 58
真菌，24，27-31，31，31n
fungi, 24, 27–31, 31, 31n
加利斯、比托里奥，268–69
Gallese, Bittorio, 268–69
赌博，144–45
gambling, 144–45
间隙连接, 37, 37
gap junctions, 37, 37
原肠胚形成，28–29, 29
gastrulation, 28–29, 29
泛化问题，126，126
generalization problem, 126, 126
生成模式（生成模型），177–81，371
generative mode (generative models), 177–81, 371
亥姆霍兹机，177–79, 178
Helmholtz machine, 177–79, 178
新皮质，181–83，188，222，258–60
neocortex as, 181–83, 188, 222, 258–60
预测一切，183–87，185
predicting everything, 183–87, 185
StyleGAN2，179，179–81
StyleGAN2, 179, 179–81
基因，18，20，304-5，363
genes, 18, 20, 304–5, 363
基因组，317
genome, 317
手势，296–97，301，310，313–14，315–16
gestures, 296–97, 301, 310, 313–14, 315–16
围棋（游戏），2，201–3，202
Go (game), 2, 201–3, 202
目标驱动行为，213–17
goal-driven behavior, 213–17
目标层次，226–31，228
goal hierarchy, 226–31, 228
古德尔，简，267–68，315–16
Goodall, Jane, 267–68, 315–16
谷歌，344
Google, 344
DeepMind，142，201
DeepMind, 142, 201
大猩猩, 239, 243, 299–300, 313
gorillas, 239, 243, 299–300, 313
八卦，337–38，339，358
gossip, 337–38, 339, 358
GPT-3（生成式预训练 Transformer 3），3–4、344–51、354–55、355
GPT-3 (Generative Pre-trained Transformer 3), 3–4, 344–51, 354–55, 355
GPT-4（生成式预训练 Transformer 4），354–56，355
GPT-4 (Generative Pre-trained Transformer 4), 354–56, 355
语法，297–98，300，336
grammar, 297–98, 300, 336
“祖母”角色
“grandmothering,” 329
颗粒前额皮质（gPFC），206，226，255-60，259，262，263，289，290，371
granular prefrontal cortex (gPFC), 206, 226, 255–60, 259, 262, 263, 289, 290, 371
颗粒细胞，206
granule cells, 206
类人猿词典，296
Great Ape Dictionary, 296
大氧化事件，21，238，374 n
Great Oxygenation Event, 21, 238, 374n
东非大裂谷，324–25
Great Rift Valley, 324–25
悲伤，59–60
grief, 59–60
杂货店购物，284–88
grocery shopping, 284–88
梳理，247，249–50，335
grooming, 247, 249–50, 335
集体生活，241–44
group living, 241–44
群体选择，333–36，337，395 n
group selection, 333–36, 337, 395n
习惯性行为，213–15
habitual behavior, 213–15
霍尔丹，JBS，334
Haldane, J. B. S., 334
幻觉，181–83
hallucinations, 181–83
哈拉瑞，尤瓦尔，303
Harari, Yuval, 303
后宫，242–44，243，388
harems, 242–44, 243, 388n
哈佛大学，97
Harvard University, 97
头部方向神经元，149
head-direction neurons, 149
希思·罗伯特 68 岁
Heath, Robert, 68
赫布，唐纳德，88岁
Hebb, Donald, 88
赫布学习，88–89，130
Hebbian learning, 88–89, 130
赫尔姆霍兹，赫尔曼·冯·赫尔姆霍兹，175–76，180–82，185
Helmholtz, Hermann von, 175–76, 180–82, 185
亥姆霍兹机，177–79, 178, 180, 182, 371
Helmholtz machine, 177–79, 178, 180, 182, 371
海洛因成瘾，230
heroin addiction, 230
后脑，95–96, 96, 149, 165
hindbrain, 95–96, 96, 149, 165
欣顿，杰弗里，6，127–28，175–77，182
Hinton, Geoffrey, 6, 127–28, 175–77, 182
海马体，149–51, 165, 190, 196, 198–99
hippocampus, 149–51, 165, 190, 196, 198–99
希波克拉底，31–32
Hippocrates, 31–32
霍布斯，托马斯，86，330
Hobbes, Thomas, 86, 330
直立人, 323, 326–30, 331–32, 341
Homo erectus, 323, 326–30, 331–32, 341
人类蜂巢思维的出现，336–41
emergence of the human hive mind, 336–41
弗洛勒斯人, 341, 341–42
Homo floresiensis, 341, 341–42
尼安德特人, 323, 331, 341, 342–43
Homo neanderthalensis, 323, 331, 341, 342–43
智人, 297, 301, 323, 331–32, 341, 342–43, 361
Homo sapiens, 297, 301, 323, 331–32, 341, 342–43, 361
马，xiv ， 223，238，239
horses, xiv, 223, 238, 239
Hubel，David，135–36，137
Hubel, David, 135–36, 137
人类增殖， 341，341–43
human proliferation, 341, 341–43
人类的独特性，295–309
human uniqueness, 295–309
尝试教猿类语言，299–301
attempts to teach apes language, 299–301
沟通，296-99
communication, 296–99
奇点，307–9
the singularity, 307–9
转移思想，301-7
transferring thoughts, 301–7
汉弗莱，尼古拉斯，239–40
Humphrey, Nicholas, 239–40
饥饿，58, 62, 79, 79–80, 119, 286, 287
hunger, 58, 62, 79, 79–80, 119, 286, 287
下丘脑，95–96, 96, 119–21
hypothalamus, 95–96, 96, 119–21
IBM 研究，108–9
IBM Research, 108–9
想法，301–2、305–6、307–8
ideas, 301–2, 305–6, 307–8
幻觉，172
illusions, 172
想象力，182–83，186–87，303
imagination, 182–83, 186–87, 303
模仿（模仿学习） ， 98–99，274–75，277–81，289–90，290，306–7
imitation (imitation learning), 98–99, 274–75, 277–81, 289–90, 290, 306–7
人工智能和，278–81
AI and, 278–81
命令式标签，297，300
imperative labels, 297, 300
归纳偏见，138，140
inductive bias, 138, 140
推理，175–77，180–82，185
inference, 175–77, 180–82, 185
抑制神经元, 38, 65
inhibitory neurons, 38, 65
内罗盘，148–49
inner compass, 148–49
内耳，124，135，140，148-49
inner ear, 124, 135, 140, 148–49
Instagram，144
Instagram, 144
意图，205，208-9，245-47，257，260
intention, 205, 208–9, 245–47, 257, 260
内部模型，146，147，151。另请参阅模型
internal models, 146, 147, 151. See also models
直觉，60–61，146
intuitions, 60–61, 146
不变性问题，133–40，134，151
invariance problem, 133–40, 134, 151
逆向强化学习，277–81
inverse reinforcement learning, 277–81
无脊椎动物，94–95，95，114n ， 116，144，151，157，237
invertebrates, 94–95, 95, 114n, 116, 144, 151, 157, 237
非自愿协会，78
involuntary associations, 78
iPhone，127
iPhone, 127
iRobot，51岁
iRobot, 51
水母、十四、27、28、29、34、38、39、43、74、80
jellyfish, xiv, 27, 28, 29, 34, 38, 39, 43, 74, 80
詹宁斯，肯，109
Jennings, Ken, 109
杰森一家（电视剧），1–2，132
Jetsons, The (TV show), 1–2, 132
约翰霍普金斯大学，131，135
Johns Hopkins University, 131, 135
约翰逊，亚当，190
Johnson, Adam, 190
共同注意，318–20，321，336，337，358
joint attention, 318–20, 321, 336, 337, 358
侏罗纪时期，162，233
Jurassic period, 162, 233
卡尼曼，丹尼尔，215
Kahneman, Daniel, 215
加拿大，86岁
Kanada, 86
埃里克·坎德尔（Eric Kandel） 76 岁
Kandel, Eric, 76
袋鼠，第十四页，223
kangaroos, xiv, 223
Kanzi（倭黑猩猩），299–300, 320
Kanzi (bonobo), 299–300, 320
卡斯帕罗夫，加里，108–9
Kasparov, Garry, 108–9
亲属选择，334–36, 337
kin selection, 334–36, 337
知识，132，246–47，257
knowledge, 132, 246–47, 257
考拉，第十四，223
koalas, xiv, 223
七鳃鳗，95，118–19，123，129
lamprey fish, 95, 118–19, 123, 129
语言，185–86、297–99、309、318–19
language, 185–86, 297–99, 309, 318–19
尝试教导猿类，299–301
attempts to teach apes, 299–301
在大脑中，310–17, 338–40, 339
in the brain, 310–17, 338–40, 339
突破＃5摘要，358，360-61
breakthrough #5 summary, 358, 360–61
人类蜂巢思维的出现，336–41
emergence of the human hive mind, 336–41
进化，302，302-3，330-33，332，358-59，360
evolution of, 302, 302–3, 330–33, 332, 358–59, 360
心理化与 353–54 之间的关系
relationship between mentalizing and, 353–54
转移思想，301-7
transferring thoughts, 301–7
语言课程，317-21
language curriculum, 317–21
大型语言模型（LLM），2-3，344-50，356-57
large language models (LLMs), 2–3, 344–50, 356–57
GPT - 3、3-4、344-51、354-55、355
GPT-3, 3–4, 344–51, 354–55, 355
GPT - 4，354–56,355
GPT-4, 354–56, 355
最后共同祖先（LUCA），19–20，24
last universal common ancestor (LUCA), 19–20, 24
晚泥盆世灭绝，158–59, 162, 238
Late Devonian extinction, 158–59, 162, 238
潜在抑制，85，85–86，90，104，195，380 n
latent inhibition, 85, 85–86, 90, 104, 195, 380n
晚二叠世灭绝事件，160–61, 237–38, 251
Late Permian extinction event, 160–61, 237–38, 251
侧皮质，149–51, 150, 165, 166
lateral cortex, 149–51, 150, 165, 166
效果律，99–100，103，144，189，213
law of effect, 99–100, 103, 144, 189, 213
第四层，172，206，206 n， 216，217
layer four, 172, 206, 206n, 216, 217
路易斯·利基，267
Leakey, Louis, 267
Leborgne，Louis Victor，310
Leborgne, Louis Victor, 310
LeCun, 扬, 10, 137 n, 186, 200, 356
LeCun, Yann, 10, 137n, 186, 200, 356
Lemoine，Blake，344
Lemoine, Blake, 344
边缘系统，8–9，9
limbic system, 8–9, 9
蜥蜴，159–60, 161
lizards, 159–60, 161
逻辑，50，185-86
logic, 50, 185–86
亮度，34–35，35 n
luminance, 34–35, 35n
说谎（骗子），334，337，396 n
lying (liars), 334, 337, 396n
猕猴, 222, 240, 243, 244, 256, 268, 313, 329, 330
macaque monkeys, 222, 240, 243, 244, 256, 268, 313, 329, 330
麦克洛斯基，迈克尔，131–32，135
McCloskey, Michael, 131–32, 135
马基雅维利式的猿猴，244–47
Machiavellian apes, 244–47
机器学习，12，84
machine learning, 12, 84
麦克莱恩，保罗，8–9，371 n
MacLean, Paul, 8–9, 371n
哺乳动物。另见特定哺乳动物
mammals. See also specific mammals
脑，95，113-14，135-36，149-50，163-66，166，186-87，203-4，205，205-7，232-33，253-55
brain, 95, 113–14, 135–36, 149–50, 163–66, 166, 186–87, 203–4, 205, 205–7, 232–33, 253–55
控制和，218-20
control and, 218–20
信用分配，195–96，196
credit assignment in, 195–96, 196
哺乳动物时代，238–39，239
Era of Mammals, 238–39, 239
进化与生命之树，xiv ， 162，163，238–39，239
evolution and tree of life, xiv, 162, 163, 238–39, 239
集体与个体之间的进化张力，241–44
evolutionary tension between the collective and the individual, 241–44
目标和习惯，213–15
goals and habits, 213–15
内在二元性，213–15
inner duality of, 213–15
做出选择，209–13
making choices, 209–13
运动皮层，223，223-26
motor cortex, 223, 223–26
运动层次，226–28, 227, 228
motor hierarchy, 226–28, 227, 228
新皮质，206–8, 207, 209, 209, 232–33, 256
neocortex, 206–8, 207, 209, 209, 232–33, 256
新皮质比率， 240，240-41
neocortex ratio, 240, 240–41
预测，184，185
prediction in, 184, 185
灵长类政治，247–52
primate politics, 247–52
模拟行动，163–64
simulating actions, 163–64
视觉皮层，135–38
visual cortex, 135–38
唯物主义，86–87
materialism, 86–87
内侧皮质。参见海马体
medial cortex. See hippocampus
模因，305
memes, 305
记忆, 76, 116
memory, 76, 116
注意力和自我控制，218–20
attention and self-control, 218–20
灾难性遗忘，131–33
catastrophic forgetting, 131–33
情节性的，196–99，232–33
episodic, 196–99, 232–33
工作，187，218，219-20
working, 187, 218, 219–20
心理化，289–91，290，361，371
mentalizing, 289–91, 290, 361, 371
突破#4摘要，289–91，360
breakthrough #4 summary, 289–91, 360
学习来源逐渐复杂化，302，302-3，360
evolution of progressively more complex sources of learning, 302, 302–3, 360
语言与 353–54 之间的关系
relationship between language and, 353–54
埃米尔·门泽尔，244–45
Menzel, Emil, 244–45
梅斯特拉尔，乔治·德，4
Mestral, George de, 4
元认知，258
metacognition, 258
小鼠，163–64, 226, 283, 296
mice, 163–64, 226, 283, 296
中脑，95–96, 96, 110, 117, 165
midbrain, 95–96, 96, 110, 117, 165
心智。参见模型；心智理论
mind. See models; theory of mind
明斯基，马文，2，103-5，120，200
Minsky, Marvin, 2, 103–5, 120, 200
镜像神经元，268–73
mirror neurons, 268–73
镜像自我认知测试，257，264
mirror self-recognition tests, 257, 264
镜像标志综合症，257–58
mirror-sign syndrome, 257–58
模型（建模）
models (modeling)
第一。看看世界上第一个模型
first. See first model of the world
第一批哺乳动物的额叶新皮质与感觉新皮质，209，209
frontal vs. sensory neocortex in first mammals, 209, 209
用心去模仿他人的心智，263–65
mind to model other minds, 263–65
其他思想，260–63，261
other minds, 260–63, 261
自己的思想，258–60，259
own mind, 258–60, 259
基于模型的强化学习，199，199–200，201–20，371
model-based reinforcement learning, 199, 199–200, 201–20, 371
AlphaZero，201-4、211、318
AlphaZero, 201–4, 211, 318
注意力、工作记忆和自我控制，218–20
attention, working memory, and self-control, 218–20
第一个目标的演变，215–17
evolution of first goal, 215–17
目标和习惯，213–15
goals and habits, 213–15
哺乳动物做出选择，209–13
mammals making choices, 209–13
预测自己，208–9
predicting oneself, 208–9
前额皮质和控制内部模拟，204-8，205，207
prefrontal cortex and controlling the inner simulation, 204–8, 205, 207
无模型强化学习，199，199–200，201，211，212，215–16，318，359–60，371
model-free reinforcement learning, 199, 199–200, 201, 211, 212, 215–16, 318, 359–60, 371
亨利·莫莱森，196–97, 198
Molaison, Henry, 196–97, 198
猫鼬，267–68，274，275
mongooses, 267–68, 274, 275
猴子，xiv， 194，247–48，269–71，284–85，287–88，316
monkeys, xiv, 194, 247–48, 269–71, 284–85, 287–88, 316
蒙塔古，里德，110，112，113
Montague, Read, 110, 112, 113
摩尔斯电码，33
Morse code, 33
动机，73–74
motivation, 73–74
运动皮层，206, 221–26, 222, 232, 241, 360
motor cortex, 206, 221–26, 222, 232, 241, 360
语言和 312
language and, 312
进化论的主要理论，222–23，223
leading theory on evolution of, 222–23, 223
镜像神经元，268–73
mirror neurons, 268–73
缺少第四层，206，206 n
missing layer four, 206, 206n
预测，223–26
predictions, 223–26
运动层次，226–31, 227, 228
motor hierarchy, 226–31, 227, 228
运动规划，224–26，270，271
motor planning, 224–26, 270, 271
芒特卡斯尔，弗农，168–70，289
Mountcastle, Vernon, 168–70, 289
多细胞生物， 24，24-26，25，28
multicellular organisms, 24, 24–26, 25, 28
多男性群体，242–44，243，387 n
multi-male groups, 242–44, 243, 387n
神话，303–4，304
myths, 303–4, 304
纳克什班迪，米里亚姆，284–85，285 n， 287–88
Naqshbandi, Miriam, 284–85, 285n, 287–88
自然选择，330，340，363
natural selection, 330, 340, 363
自然与智慧，4-6
nature and intelligence, 4–6
'nduh， 59
‘nduh, 59
负价神经元，53–55, 54, 56–57, 61, 100
negative-valence neurons, 53–55, 54, 56–57, 61, 100
线虫，xiv，46–48，47，94，101，147
nematodes, xiv, 46–48, 47, 94, 101, 147
情感状态，61–64，62，63
affective states of, 61–64, 62, 63
多巴胺和血清素，64–69, 66, 114
dopamine and serotonin, 64–69, 66, 114
权衡问题，55–57，56
problem of trade-offs, 55–57, 56
转向, 46–49, 48, 49, 53–54, 54
steering, 46–49, 48, 49, 53–54, 54
压力，69–71，73–74
stress, 69–71, 73–74
时间差异学习，115–16，116 n
temporal difference learning, 115–16, 116n
调整事物的好坏，79，79–80
tweaking goodness and badness of things, 79, 79–80
价数和, 52–55, 54
valence and, 52–55, 54
新皮质，8–9，9
neocortex, 8–9, 9
解剖学，167–72，168，205 。另请参阅无颗粒前额皮质；皮质柱；颗粒前额皮质；运动皮质
anatomy, 167–72, 168, 205. See also agranular prefrontal cortex; cortical columns; granular prefrontal cortex; motor cortex
反事实学习，192–96，193
counterfactual learning, 192–96, 193
情景记忆，196–99
episodic memory, 196–99
进化，163–64, 165–66, 166, 188, 289–90
evolution, 163–64, 165–66, 166, 188, 289–90
功能，218–20、289–90
functions, 218–20, 289–90
作为生成模型，181–83、188、222、258–60
as a generative model, 181–83, 188, 222, 258–60
语言和，312–17，315
language and, 312–17, 315
层，169，171–72，172
layers, 169, 171–72, 172
麦克莱恩的三位一体大脑假说，8–9，9
MacLean’s triune brain hypothesis, 8–9, 9
早期灵长类动物的新皮层区域，255–56，256
new neocortical regions of early primates, 255–56, 256
灵长类动物的新区域，255–56, 256, 263–64
new regions in primates, 255–56, 256, 263–64
感知，172–75
perception, 172–75
预测，183–87，185
prediction, 183–87, 185
比率， 240，240-41
ratio, 240, 240–41
感觉。参见感觉新皮质
sensory. See sensory neocortex
术语的使用，167 n
use of term, 167n
替代性试错法，189–92
vicarious trial and error, 189–92
新皮质柱。参见皮质柱
neocortical columns. See cortical columns
裙带关系，252
nepotism, 252
神经， 32
nerves, 32
神经系统，26–27，32
nervous system, 26–27, 32
神经质， 32
nervus, 32
神经回路，38–39, 39, 56, 86, 90
neural circuits, 38–39, 39, 56, 86, 90
神经游戏，109
Neurogammon, 109
神经调节剂，64–69, 66, 70–72, 71, 88, 165, 359, 371。另请参阅特定神经调节剂
neuromodulators, 64–69, 66, 70–72, 71, 88, 165, 359, 371. See also specific neuromodulators
在第一批两侧对称动物的情感状态中的作用，65–67, 66
role in affective states of first bilaterians, 65–67, 66
神经元，5,7,19,26,26-27,28-29,31-32
neurons, 5, 7, 19, 26, 26–27, 28–29, 31–32
阿德里安的发现，32–37，33
Adrian’s discoveries, 32–37, 33
皮质柱，168–72, 169
cortical column, 168–72, 169
神经科学史，31–39
history of neuroscience, 31–39
负价，53–55, 54, 56–57, 61
negative-valence, 53–55, 54, 56–57, 61
正价态，53，54，56-57
positive-valence, 53, 54, 56–57
多巴胺对预测线索、奖励和遗漏的反应，110–14，112
response of dopamine to predictive cues, rewards, and omissions, 110–14, 112
神经递质，37–38，87
neurotransmitters, 37–38, 87
艾萨克·牛顿 32 岁
Newton, Isaac, 32
纽约大学（NYU），283–84
New York University (NYU), 283–84
Ng, Andrew, 279–80
Ng, Andrew, 279–80
NMDA 受体，88 n
NMDA receptors, 88n
诺贝尔，阿尔弗雷德，76 岁
Nobel, Alfred, 76
诺贝尔奖，32，37，76
Nobel Prize, 32, 37, 76
夜间，238
nocturnal, 238
非联想学习， 80 n
nonassociative learning, 80n
去甲肾上腺素，70，123，377-78
norepinephrine, 70, 123, 377–78n
观察学习，272–77,275,280–81,306,360
observational learning, 272–77, 275, 280–81, 306, 360
“产科困境”，329
“obstetric dilemma,” 329
章鱼胺, 70, 377 n
octopamine, 70, 377n
章鱼， xiv ， 14，15，157，267–68，275，364
octopuses, xiv, 14, 15, 157, 267–68, 275, 364
奥杜威工具，326–27，327
Oldowan tools, 326–27, 327
嗅觉神经元，123–30，124，129，135
olfactory neurons, 123–30, 124, 129, 135
扩展和稀疏性，129–30，130
expansion and sparsity, 129–30, 130
嗅觉受体，123–24, 124, 381
olfactory receptors, 123–24, 124, 381n
感知的一次性特性，173–74，174
one-at-a-time property of perception, 173–74, 174
《物种起源》（达尔文），7，330
On the Origin of Species (Darwin), 7, 330
OpenAI，132，354，355，356
OpenAI, 132, 354, 355, 356
阿片类药物，70–72, 71, 74
opioids, 70–72, 71, 74
对生拇指，238
opposable thumbs, 238
情绪的起源。参见情绪、情绪的起源
origin of emotion. See emotion, origin of
生命起源，17-22
origins of life, 17–22
正交化，130
orthogonalization, 130
遮蔽，85，85–86，90，104，195，371
overshadowing, 85, 85–86, 90, 104, 195, 371
氧气，21，27
oxygen, 21, 27
氧气大屠杀，21
Oxygen Holocaust, 21
配对哺乳动物，242–44，243，329
pair-bonding mammals, 242–44, 243, 329
回形针问题，352–53
paper-clip problem, 352–53
寄生策略， 28 n
parasitic strategy, 28n
帕金森病，118
Parkinson’s disease, 118
模式识别，122–41，165
pattern recognition, 122–41, 165
灾难性遗忘，131–33
catastrophic forgetting, 131–33
计算机和 127–28
computers and, 127–28
皮质，129，129-31
cortex, 129, 129–31
歧视问题，125–26，126
discrimination problem, 125–26, 126
泛化问题，126，126
generalization problem, 126, 126
不变性问题，133–40，134
invariance problem, 133–40, 134
识别气味的问题，123–26
problem of recognizing a smell, 123–26
模式分离，130，133
pattern separation, 130, 133
伊万·巴甫洛夫, 76–79, 80, 82, 85–86, 98
Pavlov, Ivan, 76–79, 80, 82, 85–86, 98
朱塞佩·迪·佩莱格里诺，268–69
Pellegrino, Giuseppe di, 268–69
感知，172–75，218
perception, 172–75, 218
174–75, 175无法忽视的财产
can’t-unsee property of, 174–75, 175
填充属性，173，173
filling-in property of, 173, 173
173–74, 174一次一个的财产
one-at-a-time property of, 173–74, 174
二叠纪， 159，160，161，162，169
Permian, 159, 160, 161, 162, 169
二叠纪-三叠纪灭绝事件，160–61，237–38，251
Permian-Triassic extinction event, 160–61, 237–38, 251
持续狩猎，328
persistence hunting, 328
吞噬作用，23–24，28
phagotrophy, 23–24, 28
光合作用，20–22, 22, 23, 24, 27
photosynthesis, 20–22, 22, 23, 24, 27
物理学，17–18，195–96，350，363
physics, 17–18, 195–96, 350, 363
史蒂文·平克 353
Pinker, Steven, 353
盾皮鱼，157
placoderms, 157
柏拉图，86，87，330
Plato, 86, 87, 330
政治权力，247–52
political power, 247–52
波默洛，迪恩，278–79
Pomerleau, Dean, 278–79
正价神经元，53，54，56-57，100，119
positive-valence neurons, 53, 54, 56–57, 100, 119
捕食，93，122–23，243
predation, 93, 122–23, 243
预测，208–13，210，223–26
predictions, 208–13, 210, 223–26
进化， 184，184-85
evolution of, 184, 184–85
运动指令和 223–26, 271
motor commands and, 223–26, 271
新皮质和183–87,209
neocortex and, 183–87, 209
奖励预测，111，113，114 n， 115，213-14
reward-prediction, 111, 113, 114n, 115, 213–14
步骤#1：触发模拟，210，210-11
step #1: triggering simulation, 210, 210–11
步骤2：模拟选项，210，211-12
step #2: simulating options, 210, 211–12
步骤#3 ：选择一个选项，210，212-13
step #3: choosing an option, 210, 212–13
预测线索，84–86，111，112，121
predictive cues, 84–86, 111, 112, 121
前额皮质，209。另请参阅无颗粒前额皮质；颗粒前额皮质
prefrontal cortex, 209. See also agranular prefrontal cortex; granular prefrontal cortex
控制内部模拟，204-8，205，207
controlling the inner simulation, 204–8, 205, 207
运动前皮层，226, 229, 230
premotor cortex, 226, 229, 230
镜像神经元，268–73
mirror neurons, 268–73
灵长类动物。另见特定灵长类动物
primates. See also specific primates
通过观察获得新技能，275–77
acquiring novel skills through observation, 275–77
预测未来的需求，285–88
anticipating future needs, 285–88
反事实学习，194–95
counterfactual learning, 194–95
生态脑假说，282–84, 290
ecological-brain hypothesis, 282–84, 290
进化与生命之树，xiv ， 238–39，239，243–44，289–91
evolution and tree of life, xiv, 238–39, 239, 243–44, 289–91
学习来源逐渐复杂化，302，302-3
evolution of progressively more complex sources of learning, 302, 302–3
用建模思维来建模他人的思维，263–65
modeling mind to model other minds, 263–65
模仿他人的思维，260–63，261
modeling other minds, 260–63, 261
塑造自己的思维，256–60，259
modeling own mind, 256–60, 259
运动皮层，206，221，222，222-23，223，268-73
motor cortex, 206, 221, 222, 222–23, 223, 268–73
新皮质，240，240-41，313-14，360
neocortex, 240, 240–41, 313–14, 360
新的皮层区域，255–56, 256, 263–64
new neocortical regions of, 255–56, 256, 263–64
技能传承，273–77，275
skill transmission, 273–77, 275
社会脑假说，239–41，282
social-brain hypothesis, 239–41, 282
社会政治，247–52，281
social politics, 247–52, 281
社会结构，242–44，243
social structures, 242–44, 243
心智理论。参见心智理论
theory of mind. See theory of mind
工具使用，267–68，273–75
tool use, 267–68, 273–75
视觉皮层，253–55, 254
visual cortex, 253–55, 254
灵长类感觉皮层（PSC），255，258–59，354，371
primate sensory cortex (PSC), 255, 258–59, 354, 371
程序记忆，197
procedural memory, 197
蛋白质，18-19
proteins, 18–19
蛋白质合成，18-19
protein synthesis, 18–19
原始对话，318–20，336–37
proto-conversations, 318–20, 336–37
原始语言，331–32，336，358
protolanguages, 331–32, 336, 358
迷幻药，65
psychedelics, 65
精神刺激，77–78
psychic stimulation, 77–78
惩罚，337–38，358，396
punishment, 337–38, 358, 396n
拼图盒, 98, 98–99, 99, 101, 103, 115, 277, 306
puzzle boxes, 98, 98–99, 99, 101, 103, 115, 277, 306
径向对称（辐射对称），43, 44, 45, 53, 54, 80
radial symmetry (radiatans), 43, 44, 45, 53, 54, 80
Ramón y Cajal，圣地亚哥，37 岁
Ramón y Cajal, Santiago, 37
速率编码, 34–37, 36, 38
rate coding, 34–37, 36, 38
老鼠，十四
rats, xiv
预测未来的需求，284–85，285 n， 287
anticipating future needs, 284–85, 285n, 287
脑，8，78，149，150，169，189-90，198-99，206，207，213-14，223，224，229
brain, 8, 78, 149, 150, 169, 189–90, 198–99, 206, 207, 213–14, 223, 224, 229
绕行任务，191–92
detour tasks, 191–92
多巴胺和快乐，66
dopamine and pleasure, 66
多巴胺和刺激，65，66–69，110
dopamine and stimulation, 65, 66–69, 110
情景记忆，198–99
episodic memory, 198–99
观察学习，274，276-77
observational learning, 274, 276–77
遗憾，193，193–94
regret in, 193, 193–94
习惯的作用，213–14
role of habits, 213–14
角色扮演，241
role of play, 241
脊髓，78，86
spinal cord, 78, 86
可变比率加固，144
variable-ratio reinforcement, 144
替代性试错法，189–90，191–92，209–10，212，220
vicarious trial and error, 189–90, 191–92, 209–10, 212, 220
重新获得，82–84, 83, 86, 90, 371
reacquisition, 82–84, 83, 86, 90, 371
互惠利他主义，335–36
reciprocal altruism, 335–36
互惠，250，252
reciprocity, 250, 252
识别。另请参阅模式识别
recognition. See also pattern recognition
镜像自我认知测试，257，264
mirror self-recognition tests, 257, 264
新皮质和182–83,188
neocortex and, 182–83, 188
识别模式，177–79,178
recognition modes, 177–79, 178
大卫·雷迪什，190, 193, 193–94
Redish, David, 190, 193, 193–94
寄存器寻址存储器，130–31
register-addressable memory, 130–31
遗憾，192，193，193-94
regrets, 192, 193, 193–94
强化学习，101-6，164-65，192-93，359-61
reinforcement learning, 101–6, 164–65, 192–93, 359–61
根据实际奖励，107–8，108
based on actual rewards, 107–8, 108
基于预期奖励的时间差异，107–8,108
based on temporal differences in expected rewards, 107–8, 108
突破#2摘要，152–53，359–60
breakthrough #2 summary, 152–53, 359–60
学习来源逐渐复杂化，302，302-3
evolution of progressively more complex sources of learning, 302, 302–3
好奇心的重要性，142–45
importance of curiosity in, 142–45
基于模型。参见基于模型的强化学习
model-based. See model-based reinforcement learning
无模型。请参阅无模型强化学习
model-free. See model-free reinforcement learning
桑代克和，96–101
Thorndike and, 96–101
救济，115–17
relief, 115–17
REM 睡眠，182，384 n
REM sleep, 182, 384n
爬行动物的大脑，8–9，9
reptile brain, 8–9, 9
爬行动物，xiv， 159–61，162，165，165 n ，296
reptiles, xiv, 159–61, 162, 165, 165n, 296
呼吸，21–23，22，27，374 n
respiration, 21–23, 22, 27, 374n
核糖体，18
ribosomes, 18
里佐拉蒂，贾科莫，268–69
Rizzolatti, Giacomo, 268–69
Roberts, William，284–85，285 n， 287–88
Roberts, William, 284–85, 285n, 287–88
机器人技术
robotics
第一个机器人，49-52
first robot, 49–52
模仿学习，278–81
imitation learning, 278–81
罗彻斯特理工学院，139
Rochester Institute of Technology, 139
石头、剪刀、布（游戏），194–95
rock, paper, scissors (game), 194–95
Roomba ，51，51–52，53，58，64
Roomba, 51, 51–52, 53, 58, 64
机器人罗西，1–2，5，51，132
Rosey the Robot, 1–2, 5, 51, 132
罗斯，斯蒂芬，279
Ross, Stephane, 279
卢梭，让·雅克，330
Rousseau, Jean-Jacques, 330
鲁梅哈特，大卫，127–28
Rumelhart, David, 127–28
反思，192–93
rumination, 192–93
蝾螈，159
salamanders, 159
索尔克研究所，110
Salk Institute, 110
Sally-Ann 检验，260–62, 261, 262, 264
Sally-Ann test, 260–62, 261, 262, 264
盐， 79，79–80，81
salt, 79, 79–80, 81
《人类简史》（哈拉瑞），303
Sapiens (Harari), 303
饱腹感, 62, 62, 63, 66, 69, 287
satiation, 62, 62, 63, 66, 69, 287
萨维奇-朗博，苏，300
Savage-Rumbaugh, Sue, 300
“自然尺度”，14
“scale of nature,” 14
Schultz， Wolfram ，111–13，112，115
Schultz, Wolfram, 111–13, 112, 115
搜索问题，200，202，203，209，211，232
search problem, 200, 202, 203, 209, 211, 232
约翰·瑟尔，303
Searle, John, 303
热力学第二定律，17-18
second law of thermodynamics, 17–18
渴望， 59–60
sehnsucht, 59–60
癫痫，196–97，198
seizures, 196–97, 198
选择性血清素再摄取抑制剂 (SSRI)，378 n
selective serotonin reuptake inhibitors (SSRIs), 378n
自我概念（自我感觉），217，264，390
self-concept (sense of self), 217, 264, 390n
自我控制，219-20
self-control, 219–20
自动驾驶汽车，278–79
self-driving cars, 278–79
自私的基因（道金斯），305
Selfish Gene, The (Dawkins), 305
自我参照， 257
self-reference, 257
自我复制，18，19
self-replication, 18, 19
半规管， 148，148-49
semicircular canals, 148, 148–49
敏化，80 n
sensitization, 80n
感觉新皮质， 197，198，205，205-6，211-13，216-17，232，258-59，371
sensory neocortex, 197, 198, 205, 205–6, 211–13, 216–17, 232, 258–59, 371
在第一种哺乳动物中，209，209
in first mammals, 209, 209
血清素、64–69、66、71–72、73、88、359、376 n、378 n
serotonin, 64–69, 66, 71–72, 73, 88, 359, 376n, 378n
查尔斯·谢灵顿 37 岁
Sherrington, Charles, 37
手语，299，311-12
sign language, 299, 311–12
模拟，163–64，361
simulation, 163–64, 361
突破#3总结，232-33，360
breakthrough #3 summary, 232–33, 360
学习来源逐渐复杂化，302，302-3，360
evolution of progressively more complex sources of learning, 302, 302–3, 360
GPT-3 和 LLMS，349–51
GPT-3 and LLMS, 349–51
目标层次，228、229-30
hierarchy of goals, 228, 229–30
做出选择，210，210-13
making choices and, 210, 210–13
存活下来，163–64
survival by, 163–64
模拟理论，263–64
simulation theory, 263–64
技能传承，273–77，275
skill transmission, 273–77, 275
斯金纳，BF，100，144
Skinner, B. F., 100, 144
睡眠，181，182
sleep, 181, 182
嗅觉，34，38，47，53–54，123–26，135。参见嗅觉受体
smell, 34, 38, 47, 53–54, 123–26, 135. See also olfactory receptors
史密斯，尼尔，312
Smith, Neil, 312
蛇，159–60，162
snakes, 159–60, 162
社会脑假说，239–41，282，290
social-brain hypothesis, 239–41, 282, 290
社会团体，241-44
social groups, 241–44
社会等级，242–44、247–52、265–66
social hierarchy, 242–44, 247–52, 265–66
社交媒体，144–45
social media, 144–45
社会投射理论，263–64，389
social projection theory, 263–64, 389n
独居哺乳动物，242–44，243
solitary mammals, 242–44, 243
“错误来源”，77
“source of error,” 77
拱肩，340
spandrels, 340
空间地图，146–48
spatial maps, 146–48
前庭感觉, 148, 148–49
vestibular sense, 148, 148–49
说话。参见语言
speaking. See language
蜘蛛，93，158，364
spiders, 93, 158, 364
脉冲（放电）频率， 33，33–36，371
spike (firing) rate, 33, 33–36, 371
自发恢复，82–84，83，86，90，371
spontaneous recovery, 82–84, 83, 86, 90, 371
松鼠猴，284–85，287–88
squirrel monkeys, 284–85, 287–88
松鼠，163，226，271
squirrels, 163, 226, 271
“挤压问题”，35–37
“squishing problem,” 35–37
转向, 46–49, 49, 57–58, 61–64, 64
steering, 46–49, 49, 57–58, 61–64, 64
两侧对称动物，46–49, 49, 52–53
bilaterians, 46–49, 49, 52–53
突破#1摘要，90，359
breakthrough #1 summary, 90, 359
Roomba，51–52，53，64
Roomba, 51–52, 53, 64
“在黑暗中航行”，64
“steer in the dark,” 64
斯坦纳，亚当，193–94
Steiner, Adam, 193–94
兴奋剂，65
stimulants, 65
刺激强度，33–34，34，36
stimulus strengths, 33–34, 34, 36
随机神经模拟强化计算器（SNARC），103-5
Stochastic Neural-Analog Reinforcement Calculator (SNARC), 103–5
压力，69–72, 71, 90
stress, 69–72, 71, 90
急性应激反应，69–72
acute stress response, 69–72
古代应力循环，71–72，72
ancient stress cycle, 71–72, 72
慢性应激反应，72–75
chronic stress response, 72–75
中风患者，171，204–5，221，222
stroke victims, 171, 204–5, 221, 222
StyleGAN2，179，179–81
StyleGAN2, 179, 179–81
提交，242–43，247–48
submission, 242–43, 247–48
Suddendorf，Thomas，284，285，286–88，392n
Suddendorf, Thomas, 284, 285, 286–88, 392n
糖，20，21–22，27–28
sugar, 20, 21–22, 27–28
超级智能（Bostrom），352
Superintelligence (Bostrom), 352
颞上沟（STS），255 n， 371
superior temporal sulcus (STS), 255n, 371
监督学习，128，176，180
supervised learning, 128, 176, 180
萨顿，理查德，105–9, 113, 118, 120, 121, 142–43, 203
Sutton, Richard, 105–9, 113, 118, 120, 121, 142–43, 203
汗水（出汗），328
sweat (sweating), 328
符号人工智能，49-51
symbolic AI, 49–51
符号，297–98、300
symbols, 297–98, 300
突触，37，37–38，87–89，88，118，371
synapses, 37, 37–38, 87–89, 88, 118, 371
系统 1 思维，215
system 1 thinking, 215
系统 2 思维，215
system 2 thinking, 215
TD-加蒙，109，110，142，201，318，380 n
TD-Gammon, 109, 110, 142, 201, 318, 380n
温度导航，54–55
temperature navigation, 54–55
时间信用分配问题，105-7，113，120，152，200，371
temporal credit assignment problem, 105–7, 113, 120, 152, 200, 371
时间差异学习（TD学习），103–21，106，142–43，152，198–99，203，371
temporal difference learning (TD learning), 103–21, 106, 142–43, 152, 198–99, 203, 371
基底神经节，117-21
basal ganglia, 117–21
解脱、失望和时机的出现，115-17
emergence of relief, disappointment, and timing, 115–17
开发-探索困境，142–43，152
exploitation-exploration dilemma, 142–43, 152
多巴胺的重大改造，110–14
grand repurposing of dopamine, 110–14
魔法引导，105–9
magical bootstrapping, 105–9
时间差异信号（TD信号），107，111-14，152，372
temporal difference signals (TD signals), 107, 111–14, 152, 372
颞顶交界处 (TPJ), 255 n, 256, 354, 372
temporoparietal junction (TPJ), 255n, 256, 354, 372
地球改造，19-22
terraforming of Earth, 19–22
特萨罗，杰拉尔德，108–9，110
Tesauro, Gerald, 108–9, 110
四足动物，159，162
tetrapods, 159, 162
丘脑，95–96, 96, 117, 133 , 134, 139–40, 172, 172, 382
thalamus, 95–96, 96, 117, 133, 134, 139–40, 172, 172, 382n
心智理论， 246–47，260–66，268，289–90，290，372
theory of mind, 246–47, 260–66, 268, 289–90, 290, 372
通过观察获得新技能，275–77
acquiring novel skills through observation, 275–77
预测未来的需求，286–87，287
anticipating future needs and, 286–87, 287
儿童发展和，264，390 n
childhood development and, 264, 390n
用建模思维来建模他人的思维，263–66
modeling mind to model other minds, 263–66
政治活动和 281
politicking and, 281
Sally-Ann 检验，260–62, 261, 262, 264
Sally-Ann test for, 260–62, 261, 262, 264
兽孔类，160–61，162
therapsids, 160–61, 162
思考，快与慢（卡尼曼），215
Thinking, Fast and Slow (Kahneman), 215
thispersondoesnotexist.com, 179, 179–80
thispersondoesnotexist.com, 179, 179–80
爱德华·桑代克，96–100, 98, 101, 110, 111, 115, 189
Thorndike, Edward, 96–100, 98, 101, 110, 111, 115, 189
索普，查克，278–79
Thorpe, Chuck, 278–79
思维转移，301-7
thought transfer, 301–7
时间知觉，173–74，174
time perception, 173–74, 174
时机，116–17，152
timing, 116–17, 152
托尔曼，爱德华，189–90，244
Tolman, Edward, 189–90, 244
工具使用, 267–68, 273–75, 284, 327–28, 358
tool use, 267–68, 273–75, 284, 327–28, 358
奥杜威工具，326–27，327
Oldowan tools, 326–27, 327
权衡，55–57，56
trade-offs, 55–57, 56
翻译，139
translation, 139
传播性，273–77，275
transmissibility, 273–77, 275
生命之树，xiv ， 23–25，24，43，45，162
tree of life, xiv, 23–25, 24, 43, 45, 162
寒武纪祖先，94–95，95
Cambrian ancestors, 94–95, 95
人类，341，341–43
humans, 341, 341–43
哺乳动物，238–39，239
mammals, 238–39, 239
神经元激活的动物，29–30，30
neuron-enabled animals, 29–30, 30
径向对称与双侧对称，44
radial vs. bilateral symmetry, 44
试错学习，99，99–100，101，103–4，110–11，142–43，152
trial-and-error learning, 99, 99–100, 101, 103–4, 110–11, 142–43, 152
替代，189–92，211，212–13，232–33，360，361
vicarious, 189–92, 211, 212–13, 232–33, 360, 361
三叠纪，162
Triassic period, 162
二叠纪-三叠纪灭绝事件，160–61，237–38，251
Permian-Triassic extinction event, 160–61, 237–38, 251
部落主义，252，364
tribalism, 252, 364
三位一体脑假说，8–9，9，373 n
triune brain hypothesis, 8–9, 9, 373n
Tsimpli，Ianthi-Maria，312
Tsimpli, Ianthi-Maria, 312
图灵，艾伦，103
Turing, Alan, 103
海龟，159–60, 162, 319
turtles, 159–60, 162, 319
霸王龙，161
Tyrannosaurus, 161
不确定性，210–11，214
uncertainty, 210–11, 214
无条件反射，78
unconditional reflexes, 78
无意识推理，175–77，180–82，185
unconscious inference, 175–77, 180–82, 185
门控, 117–18, 120, 381 n
ungating, 117–18, 120, 381n
伦敦大学学院，216
University College London, 216
加州大学伯克利分校，189
University of California, Berkeley, 189
加州大学圣地亚哥分校，317
University of California, San Diego, 317
马萨诸塞大学阿默斯特分校，105-6
University of Massachusetts Amherst, 105–6
密歇根大学，67
University of Michigan, 67
明尼苏达大学，190
University of Minnesota, 190
帕尔马大学，268–69
University of Parma, 268–69
西安大略大学，284
University of Western Ontario, 284
无监督学习，176
unsupervised learning, 176
利用行为，229
utilization behavior, 229
V1（视觉区 1），135–37, 136
V1 (visual area 1), 135–37, 136
V2（视觉区域 2），136，136
V2 (visual area 2), 136, 136
V4（视觉区 4），136，136
V4 (visual area 4), 136, 136
价数, 4, 52–59, 54, 90, 119, 372
valence, 4, 52–59, 54, 90, 119, 372
可变比率加固，144
variable-ratio reinforcement, 144
尼龙搭扣，4
Velcro, 4
腹侧皮质，149–51, 150, 165, 166
ventral cortex, 149–51, 150, 165, 166
捕蝇草，30 n
Venus flytraps, 30n
脊柱，94
vertebral column, 94
脊椎动物。另见特定脊椎动物
vertebrates. See also specific vertebrates
脑，94–96, 97, 110–11, 118–19, 120, 120–21, 122, 129, 129, 132–33, 139–40, 140, 149, 153, 164–66, 259
brain, 94–96, 97, 110–11, 118–19, 120, 120–21, 122, 129, 129, 132–33, 139–40, 140, 149, 153, 164–66, 259
皮质，129，129-31，149-51，151，164-66，166
cortex, 129, 129–31, 149–51, 151, 164–66, 166
信用分配，195–96，196
credit assignment in, 195–96, 196
进化与生命之树，xiv ， 94–96，95，96，100，162，233，360
evolution and tree of life, xiv, 94–96, 95, 96, 100, 162, 233, 360
他们如何认识事物，124–25，125
how they recognized things, 124–25, 125
预测，184，185
prediction in, 184, 185
嗅觉和鼻子，123–26，124
smell and nose of, 123–26, 124
时间差异学习，110，114，115-16，118-19，143-44，192-93，194
temporal difference learning, 110, 114, 115–16, 118–19, 143–44, 192–93, 194
黑长尾猴, 247–48, 297, 301–2, 335
vervet monkeys, 247–48, 297, 301–2, 335
前庭感觉, 148, 148–49
vestibular sense, 148, 148–49
替代性试错法，189–92，211，212–13，232–33，360，361
vicarious trial and error, 189–92, 211, 212–13, 232–33, 360, 361
视觉，34–35，124，172–75
vision, 34–35, 124, 172–75
不变性问题，133–40，134
invariance problem, 133–40, 134
视觉皮层，134–37, 139, 167, 170, 253–55, 254
visual cortex, 134–37, 139, 167, 170, 253–55, 254
皮质柱，168–72, 169
cortical column, 168–72, 169
意志，216
volition, 216
“睡眠-觉醒算法”，182
“wake-sleep algorithm,” 182
华莱士，阿尔弗雷德，330–332
Wallace, Alfred, 330–32
想要， 68–69，111，114，114n
wanting, 68–69, 111, 114, 114n
温血，160–61，160 n， 162，163，164–65
warm-bloodedness, 160–61, 160n, 162, 163, 164–65
沃什伯恩，舍伍德，329
Washburn, Sherwood, 329
韦尼克，卡尔，311
Wernicke, Carl, 311
韦尼克区，311，311-12，313-14，316，320-21，354
Wernicke’s area, 311, 311–12, 313–14, 316, 320–21, 354
鲸鱼， xiv ， 238，239，322
whales, xiv, 238, 239, 322
托斯顿·威塞尔，135–36, 137
Wiesel, Torsten, 135–36, 137
角马，334
wildebeests, 334
威廉姆斯，罗纳德，127–28
Williams, Ronald, 127–28
意志力，219–20
willpower, 219–20
工作记忆，187，218，219-20
working memory, 187, 218, 219–20
世界模型，186，200，209，232
world models, 186, 200, 209, 232
兰厄姆，理查德，328
Wrangham, Richard, 328
濑鱼，268
wrasses fish, 268
关于作者
About the Author
MAX BENNETT是位于纽约市的 AI 公司 Alby 的联合创始人兼首席执行官。Max 之前是 Bluecore 的联合创始人兼首席产品官，该公司利用 AI 帮助世界上一些最大的品牌个性化营销。Bluecore 最近的估值超过 10 亿美元，多次跻身美国 Inc 500 家增长最快的公司之列，并入选 2018 年 Glassdoor 最佳工作场所榜单。在加入 Bluecore 之前，Max 曾在高盛担任交易员。Max 拥有多项 AI 相关技术专利，并在同行评审期刊上发表了多篇关于进化神经科学和智能主题的研究论文。Max 以优异成绩毕业于圣路易斯华盛顿大学，获得经济学和数学学士学位，并获得了约翰·M·奥林最佳经济学荣誉论文奖。2016 年，Max 入选福布斯30 位 30 岁以下精英榜单。马克斯与妻子悉尼以及他们的狗查理住在纽约布鲁克林。
MAX BENNETT is the cofounder and CEO of Alby, an AI company based in New York City. Max was previously the cofounder and chief product officer of Bluecore, a company that uses AI to help some of the largest brands in the world personalize their marketing. Bluecore was recently valued at over $1 billion, ranked numerous times on the Inc 500 fastest-growing companies in the US, and made the 2018 Glassdoor Best Places to Work list. Prior to Bluecore, Max was a trader at Goldman Sachs. Max holds several patents for AI-related technologies, and has published numerous research papers in peer reviewed journals on the topics of evolutionary neuroscience and intelligence. Max graduated from Washington University in St. Louis, summa cum laude, with a BA in economics and mathematics, where he won the John M. Olin prize for the best economics honors thesis. In 2016, Max was featured on the Forbes 30 under 30 list. Max lives in Brooklyn, New York, with his wife, Sydney, and their dog, Charlie.
在hc.com上发现伟大的作家、独家优惠等等。
Discover great authors, exclusive offers, and more at hc.com.
《情报简史》获得赞誉
Praise for A Brief History of Intelligence
“这本书揭示了你一直想知道的关于大脑的一切（但不敢问）。这是一个令人难以置信的资源。它将过去一个世纪神经科学的每一个发现融入了一个精心制作的进化叙述中。接下来的故事展示了大脑是如何从古老的蠕虫逐渐发展到我们现在这种有思想、好奇的生物的。这种综合效果很好。它的连贯性掩盖了这种处理方式几乎百科全书式的覆盖范围。”
“This book discloses everything you always wanted to know about the brain (but were afraid to ask). It is an incredible resource. It assimilates every discovery in neuroscience—over the last century—within a beautifully crafted evolutionary narrative. The ensuing story shows how an incremental elaboration of brains can be traced from ancient worms to the mindful, curious creatures we have become. The synthesis works perfectly. Its coherence obscures the almost encyclopedic reach of this treatment.”
— 卡尔·弗里斯顿 (Karl Friston)，伦敦大学学院，全球被引用次数最多的神经科学家
—Karl Friston, University College London, the #1 most-cited neuroscientist in the world
“马克斯·贝内特发表了两篇关于大脑进化的科学论文，让我大吃一惊。现在他把这些论文改编成了一本精彩的书，《智力简史》。他亲切的文笔、清晰、通俗易懂的文字以及丰富的信息量使这本书成为一本畅销书。”
“Max Bennett published two scientific papers on brain evolution that blew me away. Now he has turned these into a fabulous book, A Brief History of Intelligence. His friendly writing style, clear jargon-free prose, and well of information make this book a winner.”
— 约瑟夫·勒杜克斯，纽约大学，畅销书《焦虑》和《我们自己的深厚历史》的作者
—Joseph LeDoux, New York University, bestselling author of Anxious and The Deep History of Ourselves
“ 《智力简史》以令人难以置信的视角，整合了最相关的科学知识，描绘了人类思维如何出现的全貌……本书内容广泛、雄心勃勃、富有启发性，但仍然严格以事实为导向，避免了毫无根据的猜测。这既是一件艺术品，也是一件科学作品……我对这个在宏大的进化框架中解释整个人性的勇敢项目印象深刻。但更让我印象深刻的是，马克斯·贝内特成功完成了这项几乎不可能完成的任务。”
“With a truly mind-boggling scope, A Brief History of Intelligence integrates the most relevant scientific knowledge to paint the big picture of how the human mind emerged. . . . This text is embracing, ambitious, and lusciously enlightening but still remains strictly orientated to the facts, and avoids unsubstantiated speculation. This is both a piece of art as well as science. . . . I am deeply impressed by this brave project of explaining entire human nature in the grand evolutionary frame. But I am even more impressed that Max Bennett succeeded in this virtually impossible task.”
—库尔特·科特沙尔，维也纳大学，2010 年奥地利年度科学家奖获得者，广受好评的《狼-狗-人》一书的作者
—Kurt Kotrschal, University of Vienna, winner of 2010 Austrian Scientist of the Year Award and author of the critically acclaimed Wolf-Dog-Human
“这本书写得热情洋溢、充满活力，充满智慧和勇气。它令人大开眼界，激发智慧……这是一位年轻人的作品以及不怀私心的新鲜思维，带着纯粹的好奇心、智慧和勇气来对待问题。从年轻学生到资深学者，每个人都会觉得它有益。”
“Written with gusto and spirit, with intellectual courage and playfulness. It is eye-opening and intellectually invigorating . . . the work of a young and fresh mind that has no axes to grind and comes to the subject with untarnished joyful curiosity, intelligence, and courage. Everyone, from young students to established academics will find it rewarding.”
— 伊娃·雅布隆卡，特拉维夫大学，《四维进化》和《敏感灵魂的进化》合著者
—Eva Jablonka, Tel Aviv University, coauthor of Evolution in Four Dimensions and The Evolution of the Sensitive Soul
“马克斯·贝内特生动地描述了大脑的进化过程以及当今大脑的工作方式。《智力简史》引人入胜、内容全面，充满了新颖的见解。”
“Max Bennett gives a lively account of how brains evolved and how the brain works today. A Brief History of Intelligence is engaging, comprehensive, and brimming with novel insights.”
— 肯特·贝里奇 (Kent Berridge)，密歇根大学心理学和神经科学教授，格劳梅耶心理学奖获得者
—Kent Berridge, professor of psychology and neuroscience at University of Michigan and winner of the Grawemeyer Award for Psychology
“如果你对耳朵中间那个三磅重的灰色物体有一点点好奇，那就读读这本书吧。马克斯·贝内特的大脑自然史有趣又富有启发性，堪称杰作——既有趣又令人耳目一新。它让我的大脑很开心。”
“If you’re in the least bit curious about that three-pound gray blob between your ears, read this book. Max Bennett’s entertaining and enlightening natural history of brains is a tour de force—as refreshing as it is entertaining. It made my brain happy.”
— 乔纳森·巴尔科姆博士，畅销书《鱼知道什么》和《超级飞蝇》的作者
—Jonathan Balcombe, PhD, bestselling author of What a Fish Knows and Super Fly
“这本书为我们提供了一段探索人类智能的精彩旅程，对我们是谁以及人类的意义有着重要意义。五项‘突破’让我们与世界互动的能力变得越来越复杂，这为故事的发展提供了一种新颖的进化结构。文笔流畅，可读性强，引人入胜。强烈推荐。”
“This book provides an exciting journey through the keys to human intelligence and has important things to say about who we are and what it means to be human. The five ‘breakthroughs’ in which the ability to interact with the world gets more and more complex provides a novel evolutionary structure that carries the story forward. Well written in a very readable and engaging style. Highly recommended.”
—A. 戴维·雷迪什，明尼苏达大学， 《大脑中的思维》和《改变我们的选择方式：新的道德科学》作者
—A. David Redish, University of Minnesota, author of The Mind within the Brain and Changing How We Choose: The New Science of Morality
“如果您有兴趣了解大脑或构建类似人类的通用人工智能，您应该阅读这本书。这是一本伪装成历史的前瞻性书籍。各种神经系统的解剖学、生理学和行为的大量细节被汇集在一个连贯的进化故事中，并在其计算环境中进行解释。阅读是一种乐趣——不要错过！”
“If you are interested in understanding brains or in building human-like general AI, you should read this book. This is a forward-looking book masquerading as history. A mind-boggling amount of details of anatomy, physiology, and behavior of a variety of nervous systems are brought together in a coherent evolutionary tale and explained in their computational contexts. It is a joy to read—don’t miss it!”
—Dileep George，DeepMind，曾任 Vicarious AI 联合创始人
—Dileep George, DeepMind, previously cofounder of Vicarious AI
版权
Copyright
情报简史。版权所有 © 2023 Max Solomon Bennett。根据国际和泛美版权公约保留所有权利。支付所需费用后，您被授予非排他性、不可转让的权利，以访问和阅读屏幕上的电子书文本。未经 HarperCollins 电子书明确书面许可，不得以任何形式或任何方式（无论是电子的还是机械的，现在已知的还是以后发明的）复制、传播、下载、反编译、反向工程或存储或引入任何信息存储和检索系统。
A BRIEF HISTORY OF INTELLIGENCE. Copyright © 2023 by Max Solomon Bennett. All rights reserved under International and Pan-American Copyright Conventions. By payment of the required fees, you have been granted the nonexclusive, nontransferable right to access and read the text of this e-book on-screen. No part of this text may be reproduced, transmitted, downloaded, decompiled, reverse-engineered, or stored in or introduced into any information storage and retrieval system, in any form or by any means, whether electronic or mechanical, now known or hereafter invented, without the express written permission of HarperCollins e-books.
第一版
FIRST EDITION
封面设计：Brian Moore
Cover design by Brian Moore
封面图片 © Getty Images
Cover images © Getty Images
美国国会图书馆出版品目錄數據
Library of Congress Cataloging-in-Publication Data
姓名：Bennett，Max S. (Max Solomon)，作家。
Names: Bennett, Max S. (Max Solomon), author.
标题：智能简史：进化、人工智能以及成就我们大脑的五大突破 / 马克斯·S·贝内特 (Max S. Bennett)。
Title: A brief history of intelligence : evolution, AI, and the five breakthroughs that made our brains / Max S. Bennett.
描述：第一版。| 纽约：Mariner Books，[2023] | 包括书目参考和索引。
Description: First edition. | New York : Mariner Books, [2023] | Includes bibliographical references and index.
标识符：LCCN 2023008730 | ISBN 9780063286344 (精装本) | ISBN 9780063286368 (ebk)
Identifiers: LCCN 2023008730 | ISBN 9780063286344 (hardback) | ISBN 9780063286368 (ebk)
主题：MESH：大脑-生理学 | 生物进化 | 人工智能 | 智能-生理学
Subjects: MESH: Brain--physiology | Biological Evolution | Artificial Intelligence | Intelligence--physiology
分类：LCC QP376 | NLM WL 300 | DDC 612.8/2--dc23/eng/20230727
Classification: LCC QP376 | NLM WL 300 | DDC 612.8/2--dc23/eng/20230727
LC 记录可在 https://lccn.loc.gov/2023008730 上查阅
LC record available at https://lccn.loc.gov/2023008730
数字版 2023 年 10 月 ISBN：978-0-06-328636-8
Digital Edition OCTOBER 2023 ISBN: 978-0-06-328636-8
版本 09042023
Version 09042023
印刷版 ISBN: 978-0-06-328634-4
Print ISBN: 978-0-06-328634-4
关于出版商
About the Publisher
澳大利亚
Australia
哈珀柯林斯出版公司澳大利亚有限公司
HarperCollins Publishers Australia Pty. Ltd.
伊丽莎白街 201 号 13 楼
Level 13, 201 Elizabeth Street
澳大利亚新南威尔士州悉尼 2000
Sydney, NSW 2000, Australia
加拿大
Canada
哈珀柯林斯出版有限公司
HarperCollins Publishers Ltd
海湾阿德莱德中心东塔
Bay Adelaide Centre, East Tower
阿德莱德街西22号，41楼
22 Adelaide Street West, 41st Floor
安大略省多伦多，M5H 4E3
Toronto, Ontario, M5H 4E3
印度
India
哈珀柯林斯印度
HarperCollins India
A75，第 57 区
A 75, Sector 57
诺伊达
Noida
北方邦 201 301
Uttar Pradesh 201 301
新西兰
New Zealand
哈珀柯林斯出版公司新西兰
HarperCollins Publishers New Zealand
阿波罗大道 63 号 D1 单元
Unit D1, 63 Apollo Drive
罗斯代尔 0632
Rosedale 0632
新西兰 奥克兰
Auckland, New Zealand
英国
United Kingdom
哈珀柯林斯出版有限公司
HarperCollins Publishers Ltd.
伦敦桥街 1 号
1 London Bridge Street
英国伦敦 SE1 9GF
London SE1 9GF, UK
美国
United States
哈珀柯林斯出版公司
HarperCollins Publishers Inc.
百老汇 195 号
195 Broadway
纽约，NY 10007
New York, NY 10007
*我让 GPT-3 完成以下句子：“我在没有窗户的地下室里，我仰望天空，我看到了……” GPT-3 说：“一盏灯，我知道那是一颗星星，我很高兴。” 事实上，如果你在地下室向上看，你不会看到星星，你会看到天花板。 2023 年发布的 GPT-4 等较新的语言模型成功地以更高的准确性回答了这类常识性问题。 请继续关注第 22 章。
* I asked GPT-3 to complete the following sentence: “I am in my windowless basement, and I look toward the sky, and I see . . .” GPT-3 said “a light, and I know it is a star, and I am happy.” In reality, if you looked upward in a basement you would not see stars, you would see the ceiling. Newer language models like GPT-4, released in 2023, successfully answer commonsense questions like this with greater accuracy. Stay tuned for chapter 22.
*尽管，就像进化过程中的一切一样，其中也存在细微差别。有些动物和真菌物种采用了第三种中间选择：寄生策略。寄生虫不会主动捕捉猎物并将其杀死，而是感染猎物并窃取糖分或从内部将其杀死。
* Although, as with everything in evolution, there is nuance. There is a third, middle option that some species of both animals and fungi settled into: the parasitic strategy. Instead of actively catching prey to kill it, parasites infect prey and steal sugar or kill them from the inside.
*当然，你也可以反过来说：真菌从来没能以其他生命为食，因为它们从来没能获得神经元。问题不在于哪个先出现，而是神经元和捕食二级多细胞生命是同一种策略的一部分，而真菌从来没用过这种策略。
* You could, of course, state this the opposite way: fungi never got to feed on other life because they never got neurons. The point is not which came first but that neurons and hunting level-two multicellular life were part of the same strategy, one which fungi never used.
*在这项实验中，研究人员证实，这种影响不仅仅是由过度接触盐引起的，而是由刺激物（盐）与饥饿的负面情感状态之间的关联引起的。研究人员让第三组线虫在盐水中呆同样长的时间，但也在盘子里加入了食物，这样线虫就不会感到饥饿。这第三组线虫接触了同样多的盐，但之后仍然很高兴地转向盐。这表明，避免接触盐并不是由过度接触盐引起的，而是由盐和饥饿之间的关联引起的（见图4.1中的中间示例）。
* In this experiment, researchers confirmed that this effect was not caused just by overexposure to salt, but instead by the association between a stimulus (salt) and the negative affective state of hunger. Researchers took a third group and made them spend the same number of hours in salt water but also added food to the dish so nematodes wouldn’t experience hunger. This third group, which experienced the same amount of exposure to salt, still happily steered toward salt afterward. This suggests the salt avoidance was not caused by overexposure to salt, but by the association between salt and hunger (see the middle example in figure 4.1).
*这些更为远亲的动物确实参与了所谓的非联想学习，例如适应（正如埃德加·阿德里安所发现的），以及另一种类似的学习，称为敏感化，即当反射响应先前引起的刺激而增强时。
* These more distant animals do engage in what is called nonassociative learning, such as adaptation (as Edgar Adrian found), and another similar type of learning called sensitization, which is when reflexes strengthen in response to a previously arousing stimulus.
*多巴胺神经元始终具有背景静态，每秒发射约一到两个脉冲。在这些缺失期间，这些神经元会保持沉默（见图6.3）。
* Dopamine neurons always have a background static, firing at about one to two spikes per second. During these omissions, these neurons go silent (see figure 6.3).
*值得注意的是，一些无脊椎动物，特别是节肢动物，确实表现出这种奖励预测错误，但考虑到这些奖励预测错误在其他简单的两侧对称动物中没有发现，并且节肢动物中发现这些反应的大脑结构是节肢动物独有的大脑结构，因此人们认为这是独立进化的。
* It is important to note that some invertebrates, specifically arthropods, do show such reward-prediction errors, but this is believed to have evolved independently given the fact that these reward-prediction errors are not found in other simple bilaterians, and the fact that, in arthropods, the brain structures these responses are found within are uniquely arthropod brain structures.
*事实上，最近的研究表明，进化巧妙地改变了多巴胺的功能，同时仍然保留了其早期产生渴望状态的作用。基底神经节输入核（称为“纹状体”）中的多巴胺含量似乎可以衡量折扣预测的未来奖励，根据事情可能有多好来触发渴望状态，并驱使动物关注和追求附近的奖励。当动物接近奖励时，多巴胺会增加，在动物期望获得奖励时达到峰值。在这个增加过程中，如果预测的奖励发生变化（某些遗漏或新线索改变了获得奖励的概率），那么多巴胺水平就会迅速增加或减少，以解释新的预测未来奖励水平。多巴胺水平的这些快速波动是通过舒尔茨发现的多巴胺神经元的爆发和暂停产生的；多巴胺水平的这些快速波动是时间差异学习信号。纹状体中多巴胺的数量会改变神经元的兴奋性，从而使行为转向利用和欲望。相反，多巴胺水平的快速变化会触发各种连接强度的变化，从而强化和惩罚行为。换句话说，脊椎动物的多巴胺既是欲望的信号，也是强化的信号。
* In fact, recent studies show how elegantly evolution modified the function of dopamine while still retaining its earlier role of generating a state of wanting. The amount of dopamine in the input nuclei of the basal ganglia (called the “striatum”) seems to measure the discounted predicted future reward, triggering the state of wanting based on how good things are likely to be and driving animals to focus on and pursue nearby rewards. As an animal approaches a reward, dopamine ramps up, peaking at the moment when an animal expects the reward to be delivered. During this ramping-up process, if predicted rewards change (some omission or new cue changes the probability of getting a reward), then dopamine levels rapidly increase or decrease to account for the new level of predicted future reward. These rapid fluctuations in dopamine levels are produced through the bursting and pausing of dopamine neurons that Schultz found; these rapid fluctuations in dopamine levels are the temporal difference learning signal. The quantity of dopamine floating around in the striatum modifies the excitability of neurons, which shifts behavior toward exploitation and wanting. In contrast, the rapid changes in dopamine levels trigger modifications in the strength of various connections, thereby reinforcing and punishing behaviors. In other words, dopamine in vertebrates is both a signal for wanting and a signal for reinforcement.
*需要一定的实验智慧才能区分仅仅因为偶然性不再适用而消失的关联（例如，光不再导致电击）和从省略某些东西中学习。在一项对鱼的研究中，通过在省略奖励的试验中特别添加新线索来显示区别。如果关联只是消失，那么这个新线索不会成为奖励（在省略试验中没有得到任何强化），但如果动物的大脑将省略的电击本身视为奖励，那么这个新线索（在省略电击时独特地出现）应该被学会作为奖励。研究人员已经表明，在这样的实验中，鱼确实将这个新线索视为奖励，并会在将来接近它。相比之下，我们知道线虫不能做到这一点，因为它们甚至不能将时间上分开的事件联系起来，有证据（尽管仍未解决）表明，即使是聪明的无脊椎动物，如蜜蜂和螃蟹，也不会以这种方式从省略中学习。
* It requires some experimental cleverness to distinguish between an association merely fading because a contingency no longer applies (e.g., a light no longer leads to a zap) and learning from the omission of something. In one study with fish, the distinction was shown by adding a new cue specifically in trials where rewards were omitted. If an association were merely fading, then this new cue would not become rewarding (nothing was reinforced in the omission trial), but if instead an animal’s brain treats an omitted zap as rewarding in and of itself, then this new cue (which showed up uniquely when zaps were omitted) should be learned to be as rewarding. Researchers have shown that in such experiments fish do, in fact, treat this new cue as rewarding and will approach it in the future. In contrast, we know a nematode cannot do this because they cannot even associate events separated in time, and there is evidence (although it is still unsettled) that even smart invertebrates such as honeybees and crabs do not learn from omission in this way.
*请注意，他没有使用“卷积”这个词，但他被认为是提出这种方法和架构的人。还请注意，是 Yann LeCun 更新了这个架构以使用反向传播，这催化了卷积神经网络在实际应用中的广泛采用。
* Note that he did not use the word convolution, but he is credited with coming up with the approach and architecture. Also note that it was Yann LeCun who updated this architecture to use backpropagation, which is what catalyzed the widespread adoption of convolutional neural networks in practical applications.
*不过，公平地说，鱼类和爬行动物的大脑确实存在差异。有人认为羊膜动物进化出了背部皮质，这可能是新皮质的前身（尽管新证据表明，早期羊膜动物并不存在背部皮质）。
* Although, to be fair, there are differences between fish and reptile brains. Some argue amniotes evolved a dorsal cortex, a possible precursor to the neocortex (although newer evidence suggests the dorsal cortex was not present in early amniotes).
*正如第 11 章所提到的，它之所以被称为颗粒状，是因为它的第四层非常厚，其中包含颗粒神经元。
* As mentioned in chapter 11, it is called granular because of its uniquely thick layer four, which contains granule neurons.
*请注意，Naqshbandi 和 Roberts 进行了初步基线实验，以确保枣的数量和葡萄干的数量在猴子和老鼠身上引起相似的口渴程度，测量方法是当动物在食用这些数量的枣或葡萄干的同时可以自由饮水时，水摄入量的相对增加量。
* Note that Naqshbandi and Roberts did an initial baseline experiment to ensure the quantity of dates and the quantity of raisins each induced similar levels of thirst in monkeys and rats, as measured by the relative increase in water intake when animals were given free access to water alongside these quantities of dates or raisins.